Critical references: The Model AI Infrastructure Authorization Code · The Data Center Authorization Market: A 50-State Regulatory Atlas

Thesis. Match each class of AI computation to the electricity and jurisdiction it can run on, and clean power carries far more of the compute the United States has reason to keep.

Why now. PJM’s 2026 large-load framework, FERC’s June 2026 large-load orders promoting flexible transmission service, and recent nuclear and geothermal offtake put firmness on the table this planning cycle. Every audience faces the placement decision now: policymakers · operators · counsel · investors.

Executive Summary

The clean-power debate for Artificial Intelligence (AI) has focused heavily on supply. Build more generation, more transmission, more storage. A second, underused lever sits on the demand side. Match each class of computation to the electricity it can run on, and the same clean generation carries far more of the compute the United States has reason to keep.

One effect governs the argument, and the paper names it the firmness inversion. Moving flexible computation toward abundant energy lowers total domestic load. The workloads left behind grow more dependent on firm power, because latency and security hold them in place. Firm clean generation becomes more valuable as flexible compute leaves, not less.

The rule the paper defends compresses to one line. Move the compute that can move, and firm the compute that cannot.

The thesis follows. Through 2040, the binding infrastructure question shifts from where to site a data center to which computation belongs where. Operators who separate workloads by firmness and mobility will run more of their anchored compute on clean power than operators who keep optimizing campus scale.

MindCast builds Cognitive Digital Twins (CDTs) of the operators, utilities, and jurisdictions in the contest. The twins simulate their decisions under pressure.

Behavioral economics supplies the decision rules, including the bigger-is-better heuristic that keeps capital flowing to campuses the grid cannot firm. Game theory supplies the payoff structure and the equilibrium the bargain reaches. Predictive Behavioral Economics + Dynamic Game Theory produces the forecast.

The paper evaluates why concentration collides with the grid, then separates compute by its power and geography needs. It locates the domestic floor the United States must firm, then defines the bargain operators and jurisdictions strike through 2040.

The formal simulation releases six MindCast Simulation Predictions, one primary and five secondary. The retained tier buys firmness at 76-88%, the primary claim. Workload differentiation becomes public strategy at 82-90%, and capacity-backed service governs constrained conditions at 84-92%.

🏛️ Policymakers should write firmness and mobility into authorization terms, and value accredited capacity rather than raw megawatts.

💼 Executives should move from campus-scale optimization to compute-portfolio optimization, and place each workload where its power lives.

⚖️ Counsel should treat offshore placement as a move inside the export-control perimeter, mapped to current destination controls rather than around them.

📊 Investors should evaluate the firmness profile of retained load, because the retained tier drives firm-clean demand more than announced gigawatts do.

I. The Gigawatt Concentration Problem

Scale built the current data center. Larger campuses lowered the cost of construction, networking, and operations. The industry made the multi-gigawatt campus its default, and the grid did not cooperate.

A multi-gigawatt load concentrates generation, transmission, and reliability demand at one point. Grid headroom is finite at any point. The larger the single load, the more of a region’s remaining capacity it consumes, and the harder its consent becomes to win.

Concentration also collides with a rule most siting analysis skips. Connection is not firmness. On July 27, 2026, the PJM Board directed development of an interim resource-adequacy framework. Under the framework, qualifying large loads without sufficient capacity backing could face earlier reductions during shortage conditions. PJM is the largest United States regional transmission organization (RTO).

A campus can clear every state gate, energize, and still face earlier curtailment if its service lacks sufficient capacity backing. Giant campuses will keep being built where the power supports them. The open question is whether the giant campus stays the default destination for every incremental workload, and the grid is making that assumption progressively harder to defend.

II. The Method: Predictive Behavioral Economics + Dynamic Game Theory

MindCast is a simulation firm built on Predictive Behavioral Economics + Dynamic Game Theory. MindCast builds Cognitive Digital Twins of the operators, utilities, and jurisdictions in a contest. The twins run forward under pressure.

Behavioral economics supplies the decision rules. Bounded rationality, salience, and loss aversion set them. An operator over-weights campus scale and under-weights firmness, and a community resists a cost that lands far from its benefit.

Dynamic Game Theory supplies the payoff structure and the equilibrium the contest reaches. The game itself changes while the actors play. Rules, rivals, and available strategies shift mid-contest. Neither discipline predicts alone, and the combination produces MindCast Foresight Simulation Predictions, each with a band and a falsifier.

The simulation runs opposing twins against each other. An operator plays a utility, and a developer plays a regulator, so the forecast reflects the countermoves each actor provokes rather than one actor’s plan in isolation.

III. Not All Compute Needs The Same Geography

Compute is not one product, and the grid treats its varieties differently. Workloads differ by latency, interruptibility, and security. Power intensity separates them further.

Production inference answers users in real time. Latency binds it near the people it serves. Training runs over days rather than in the moment, so it follows energy more readily than latency-bound inference, within the limits utilization and synchronization set. The largest frontier runs are the exception, because tight low-latency fabric across tens of thousands of accelerators keeps them anchored in the retained tier. Batch inference and preprocessing fall between the two.

Inference and flexible compute split cleanly. Inference tends to move toward users. Flexible compute can move toward energy. A firm that sorts its workloads this way gains a placement it can defend to a utility, a regulator, and a community.

Workload class should become a primary determinant of where the next megawatt of AI load lands.

IV. Connected Is Not Firm

Firmness is the property the clean-power debate keeps collapsing. A facility can hold interconnection, buy annual clean energy, and still fail to deliver power when the servers run. Three products hide inside the word firm, and the weakest one governs.

The firmness equation states it plainly.

Firm Service = minimum ( Interconnection, Deliverable Energy, Accredited Capacity )

Interconnection is permission to attach. Deliverable energy is power that reaches the site over available transmission. Accredited capacity is supply the grid counts on during a shortage. A site missing any one of the three stays exposed, whatever the other two promise.

Clean supply divides on the same line. Annual matching proves procurement over a year. Hourly matching aligns purchased generation with consumption hour by hour, and still does not prove nodal deliverability or capacity.

Firm or highly dependable clean resources can provide stronger capacity contribution than standalone variable generation, depending on configuration and accreditation. Variable clean resources earn partial capacity credit. Storage and complementary supply raise portfolio capacity contribution, and transmission improves deliverability.

A clean mandate written without the capacity layer can produce sites that are green on paper yet exposed to earlier curtailment under constrained conditions. Policymakers hold that risk directly.

V. The Domestic Compute Floor

Some computation carries enough value in American geography to stay regardless of where power runs cleanest. The paper calls that set the domestic compute floor. The floor holds the workloads the United States has economic or strategic reason to keep onshore.

Production inference leads the set, held by latency to the users it serves. Four further categories stand as candidates the formal run resolves. Defense and critical-infrastructure compute qualify on security, and regulated data qualifies on law. Sensitive model operations and selected frontier training qualify on control and strategic value.

The floor changes what the United States must generate. Move flexible compute toward abundant energy, and total domestic load falls against the counterfactual. The load that remains is the load hardest to move, and hardest to run on intermittent supply.

The retention boundary is contested, and the paper holds it open rather than fixing it in advance. Define the floor first, because the floor sets the firm-power bill, not the announced gigawatt total.

VI. The Firmness Inversion

The retained floor produces the paper’s central and least intuitive finding. Compartmentalization does not relieve the American clean-firm problem. Compartmentalization concentrates it.

The firmness inversion works in two steps. First, flexible workloads leave for energy-advantaged sites at home or abroad, and aggregate domestic load falls. Second, the workloads that remain are latency-bound and security-bound, so their firmness requirement per megawatt rises. Total load falls, and the remaining load runs firmer.

Firmness rises on two axes at once. Retained inference needs firm service to run around the clock. Retained inference also needs a capacity-backed service arrangement where the governing market differentiates shortage treatment by capacity support. A site can contract clean power and still fail the capacity test.

The inversion favors firm clean resources. Nuclear and hydro and geothermal gain strategic value precisely as flexible compute departs. Variable clean generation gains a complementary role, paired with the flexible workloads that bend to it. Moving flexible compute away raises the firm-clean intensity of what stays, so specialization makes firm clean more valuable, not less.

VII. Training Leads Per Site, Inference Leads The Fleet

Compute power ranks two ways, and the two rankings invert. By power density at a single site, model training leads. By total energy across the deployed fleet, inference leads. A training run is enormous and rare, while an inference query is small and constant, so each measure crowns a different workload.

Density ranks by what strains one grid connection. A frontier pretraining run tops it, executed as a single synchronized job across thousands of graphics processing units (GPUs), with a single run projected to reach 4 to 16 GW of peak power by 2030. Fine-tuning follows on the same continuous profile at lower magnitude. Image and video generation, real-time text inference, and batch work rank below it.

Fleet energy ranks by what the industry actually burns. Real-time inference tops it, at 80-90% of total AI energy, because query volume swamps the one-time cost of training. Pretraining and fine-tuning make up most of the rest, concentrated and spiky rather than continuous. Image and video generation form a rising and poorly-measured slice, since operators seldom disclose per-task energy.

Cooling and facility overhead sit across both rankings. Non-compute overhead adds roughly 10 to 40 percent on top of the servers, so it outweighs several workload classes on its own. The rankings matter to placement because they cross: training leads on density and can move to abundant energy, while inference leads on fleet energy and stays near users. Retaining the fleet’s largest load concentrates the firmest power requirement at home, which is the firmness inversion measured in consumption.

VIII. Matching Workloads To Power

Placement becomes an allocation problem once workloads and resources are sorted. Each compute class carries a firmness need and a geography need. Each resource carries a delivery shape. Good placement pairs them.

Persistent, high-utilization workloads suit firm resources, so nuclear and hydro and geothermal anchor them where available. Flexible workloads suit variable resources, so wind and solar with storage carry training that follows the curve. Transmission opens geographic arbitrage and adds its own constrained layer.

Compute can also provide grid value rather than sit as a passive load. A training cluster that ramps with wind, or pauses during a shortage, delivers demand response the grid would otherwise procure. Flexibility earns capacity-equivalent value only where the market accredits it, and Duke research estimates roughly 98 GW of new load could be integrated across US balancing authorities if flexible loads accept an average annual curtailment rate of 0.5%. Interruptibility converts a liability into a service.

The right site pairs a workload’s firmness need with a resource’s delivery shape, and lets flexible compute earn capacity-equivalent grid value.

IX. Energy-Advantaged And Offshore Placement

Flexible compute follows energy, and the first move is domestic. Pacific Northwest hydro, wind-rich interior grids, and geothermal basins offer clean power away from the metros that host inference demand. A training campus does not have to leave the country to leave a congested grid.

Offshore placement extends the same logic across a border, and complements domestic capacity rather than replacing it. Allied clean-power jurisdictions can host energy-intensive training with limited user-proximity requirements. Canadian hydro and Nordic firm renewables lead the set. Offshore capacity works as a release valve that preserves scarce domestic firm power for the anchored floor.

Offshore access opens only inside a legal boundary, and Counsel should treat the boundary as the design constraint. United States export-control law gates advanced computing hardware and advanced AI model weights by destination, and separates allied jurisdictions from adversary ones. License Exception Artificial Intelligence Authorization and Supplement No. 5 to Part 740 name the eligible allied jurisdictions, and both remain current law amended through 2026 as enforcement posture shifts.

Four factors decide offshore eligibility together, not a frontier line. Workload mobility and jurisdiction eligibility set it, and security classification and infrastructure economics complete it. Geographically mobile workloads are the strongest initial candidates, and the permitted set can widen or narrow on a rule change no operator controls.

X. The Grid-Anchored Bargain Through 2040

Authorization, energy, and compute architecture converge into one bargaining system. Utilities and jurisdictions hold generation, transmission, and authorization. Operators hold flexible load, curtailment capability, and generation investment. Each side wants what the other controls.

Game theory sets the payoff structure. Jurisdictions weigh consent, firm supply, and ratepayer protection. Operators weigh speed, certainty, and firm service. Standardized terms turn the contest into a competition for scarce authorization, and the operator who brings flexibility and firm-clean investment clears terms the announce-first operator cannot.

Behavioral economics explains the friction. The bigger-is-better heuristic keeps capital flowing toward campuses the grid cannot firm, because scale is salient and firmness is not. Communities weigh concentrated local benefit against diffuse ratepayer cost, and loss aversion sharpens resistance when the bill lands far from the jobs. Executives who read those decision rules bargain better than executives who model only megawatts.

The simulation resolves the bargain toward a portfolio of specialized sites anchored to different power systems, not a single optimized campus. The long-horizon question follows. Does the industry optimize the next generation of AI infrastructure by campus, or by a portfolio of specialized compute anchored to the grids that firm each workload?

XI. MindCast Simulation Predictions

The paper turns on one question. Can workload architecture increase the amount of strategically important AI that clean power can firm? The simulation releases six Simulation Predictions, one Primary (P) and five Secondary (S). Each carries a band, a falsifier, and a public source.

Major operator is frozen to seven firms for validation: Microsoft · Google · Amazon · Meta · OpenAI · Anthropic · xAI. Primary marks causal weight, not the highest band.

Competing hypothesis. Supply-side and market reform could preserve the economic advantage of heterogeneous mega-campuses long enough that workload-by-geography separation stays marginal through 2029. Contrary-route test: fewer than two major operators adopt explicit workload-by-geography strategies while new large campuses keep absorbing heterogeneous workload classes without differentiated-service constraints.

P1 · The retained tier buys firmness (76-88%): Two or more major operators pair anchored United States compute with firm-clean or capacity-backed supply, distinct from their flexible workloads. Validated against power purchase agreement (PPA) and interconnection disclosures by December 31, 2029. Falsifier: fewer than two major operators publicly pair anchored compute with firm-clean or capacity-backed supply by the window.

S1 · Workload differentiation becomes public strategy (82-90%): Two or more major operators publicly separate workloads by firmness and geographic mobility as infrastructure strategy. Validated against operator infrastructure disclosures and earnings materials by December 31, 2029. Falsifier: operators keep guiding toward undifferentiated campus scale with no public placement strategy.

S2 · Flexible compute relocates toward energy, domestic first (72-81%): At least three major new flexible AI compute sites locate in energy-advantaged domestic regions rather than the demand metros. Validated against site announcements and interconnection filings by December 31, 2029. Falsifier: fewer than three such sites locate outside the demand metros by the window.

S3 · Firm clean gains strategic value (78-88%): Cumulative disclosed firm-clean nameplate capacity contracted by AI operators, in megawatts, grows faster than cumulative disclosed variable-only nameplate capacity contracted by those operators. Validated against procurement disclosures by December 31, 2029. Falsifier: firm-clean nameplate contracted stays flat or declines against variable nameplate.

S4 · Offshore compute complements domestic capacity inside the permitted perimeter (60-73%): At least one major operator places a substantial energy-intensive, geographically mobile AI workload in an eligible allied jurisdiction under United States export-control compliance. Validated against operator announcements and regulatory filings by December 31, 2030. Falsifier: no such placement, or placement outside the eligible perimeter. Prediction-break: a federal rule change that materially widens or narrows the eligible set.

S5 · Capacity-backed service governs constrained conditions (84-92%): At least one major RTO or large-load regime operationalizes service or curtailment differentiation by capacity backing, firm supply, or enforceable flexibility. Validated against RTO filings and operative tariffs by December 31, 2028. Falsifier: no such differentiation is operational in any major market by the window.

Two interpretive findings sit outside the event pool, and do not pool with the bands above.

Firmness inversion (85-92% directional): Moving mobile compute away lowers aggregate domestic demand while raising the firmness intensity of the load that remains. The mechanism is the paper’s governing contribution.

Portfolio advantage (directional, held moderate-high): Operators able to allocate workloads across differentiated sites gain greater clean-power optionality than operators tied to concentrated load. The advantage is structural, not exclusive, because a concentrated operator can still specialize over time or contract firm supply.

The primary prediction carries the paper’s weight, and the firmness inversion is the mechanism beneath every entry.

XII. Risk Mitigation

Each prediction converts into a stakeholder exposure and a set of unilateral actions. Per prediction the layer names the loss on inaction, the unilateral moves that reduce it, and the residual that survives full mitigation. Actions are analytic options rather than recommendations, and none assumes advance knowledge of the outcome.

P1 · The retained tier buys firmness (76-88%).

💼 Executive exposure: an operator that anchors latency-bound compute without firm clean or accredited capacity can face higher power costs, curtailment exposure, and delayed time-to-compute on the retained fleet. Actions: energy procurement contracts firm clean or accredited capacity for anchored sites by the 2029 checkpoint; operations secures qualifying capacity-backed arrangements ahead of the resource-adequacy window; portfolio planning classifies the fleet by firmness need now. Residual: firm-clean lead times exceed the contracting window, so a share of anchored load runs unfirmed through the buildout.

S1 · Workload differentiation becomes public strategy (82-90%).

💼 Executive exposure: an operator optimizing undifferentiated campus scale while rivals differentiate loses the marginal siting decision and strands interconnection deposits at congested sites. Actions: strategy formalizes a workload-placement architecture separating classes by firmness and mobility by the 2029 checkpoint; siting redirects the next incremental campus to the differentiated rule. Residual: sunk mega-campus commitments cannot be re-differentiated, so legacy concentration keeps its firmness cost.

S2 · Flexible compute relocates toward energy, domestic first (72-81%).

🏛️ Policymaker exposure: an energy-advantaged state that does not ready interconnection and firm supply forgoes the construction spend and tax base a relocation wave brings, measured against the session calendar. Actions: the legislature writes a large-flexible-load authorization track with firm-supply and interconnection terms by the next session; the utility commission pre-clears a flexible-load tariff so terms sit ready. Residual: transmission buildout exceeds a session cycle, so a state can write terms and still lack deliverable headroom.

💼 Executive exposure: an operator anchoring flexible training in a demand metro can pay for locational firmness beyond the workload’s geographic requirement, measured in power cost per megawatt-hour above an energy-advantaged site. Actions: siting routes new flexible workloads to energy-advantaged regions; network engineering provisions backhaul so relocation holds the workload together. Residual: some flexible workloads carry data-gravity ties to a metro that relocation cannot fully sever.

S3 · Firm clean gains strategic value (78-88%).

💼 Executive exposure: an operator without secured firm-clean offtake competes for a shrinking pool as rivals contract it, measured in the premium paid for late procurement. Actions: procurement weighs earlier firm-clean contracting or equity participation where it reduces supply risk; corporate development evaluates development partnerships with advanced nuclear or geothermal projects. Residual: firm-clean deployment timelines cap how fast contracted supply becomes delivered supply.

S4 · Offshore compute complements domestic capacity inside the permitted perimeter (60-73%).

⚖️ Counsel exposure: an operator placing compute offshore without mapping current export-control destination rules risks a violation, measured in licensing delay and enforcement exposure. Actions: trade compliance classifies each offshore-candidate workload against current Export Administration Regulations destination controls and license requirements before deployment; legal builds a jurisdiction-eligibility gate on offshore siting. Residual: export-control policy can shift within the window, so a compliant placement today can fall outside the eligible set faster than an operator can relocate the load. Constraint: counsel cannot pre-clear an export with the Bureau unilaterally, so the action is internal classification and a license application, not an assurance of approval.

💼 Executive exposure: an operator treating offshore as a substitute over-commits abroad and under-firms the retained domestic tier, measured in quarters of retained-tier firmness deferred by the offshore bet. Actions: infrastructure strategy sizes offshore to the mobile tier only; portfolio planning holds domestic firm-clean procurement on its own track. Residual: allied-host transmission and interconnect limits cap how much mobile load offshore can absorb.

S5 · Capacity-backed service governs constrained conditions (84-92%).

🏛️ Policymaker exposure: a market without capacity-backing and flexibility terms before the next shortage faces contested curtailment events it cannot adjudicate cleanly, measured in disputes per constrained season. Actions: an RTO files capacity-accreditation and flexibility-service rules with FERC ahead of the constrained season; a state commission acts through its own authority to set large-load service terms. Residual: FERC review timing sits outside the filer’s control, so a filed rule can lag the shortage it governs. Constraint: an RTO rule requires FERC action to take effect, so the unilateral move is the filing and the interim tariff, not the approval.

💼 Executive exposure: an operator whose large load lacks capacity backing can face earlier curtailment under applicable capacity-linked service rules, measured in hours of forced curtailment per constrained season. Actions: operations secures accredited capacity or enrolls enforceable flexibility for constrained-condition service; engineering builds curtailment-tolerant scheduling so a shortage lowers throughput rather than halting it. Residual: a severe multi-day shortage exceeds what enrolled flexibility can absorb, so residual curtailment survives on the retained tier.

Several actions here, firm-clean procurement and workload differentiation among them, also reduce exposure in the buy-side bargaining companion. The linkage holds, and neither analysis claims the action twice.

XIII. What To Watch

Three observables decide the direction before the formal windows close. PJM’s Interim Resource Adequacy Service reaches an operative tariff or stalls at FERC. Hyperscaler firm-clean offtake, in nuclear and geothermal, either accelerates or plateaus through 2027. New flexible-compute sites announce in energy-advantaged regions or concentrate in the demand metros.

The dominant fork is federal. Export-control posture on allied-jurisdiction compute can widen or narrow the offshore option, and the administration’s move away from a single replacement rule leaves the eligible set open to amendment. Watch the rulemaking that follows the 2026 amendments, because it sets whether the mobile tier can leave.

XIV. Conclusion

Scale built the data center, and firmness is unbuilding the mega-campus as the universal default. Grid rules increasingly differentiate how large loads can be served, and the workloads that cannot move inherit the firmest, scarcest power. Operators who match computation to electricity and jurisdiction will run more of their anchored compute on clean supply than operators who keep optimizing campus scale.

The paper releases six MindCast Simulation Predictions on that mechanism, led by the retained tier buying firmness at 76-88%. Move the compute that can move, and firm the compute that cannot.

Appendix: MindCast Corpus

The paper builds on prior MindCast work. Each entry carries its role in one line.

The AI Data Center Authorization Price: A 50-State Baseline (2026). The sell-side baseline this paper’s demand architecture bargains against.

AI Data Center Authorization Bargaining Power (2026). The buy-side operator ratings whose clean-portfolio axis this paper deepens.

The Authorization Market (2026). The series umbrella and the four-layer authorization stack this paper sits inside.

The Model Data Center Authorization Code (2026). The measurement instrument behind the firmness and clean-energy provisions.

The Power Stack (2025). The resource hierarchy this paper extends into workload placement.

AI Datacenter Edge Computing: Ship the Workload Not the Power (2025). The origin of the ship-the-workload principle carried here into the firmness frame.

VRFB’s Role in AI Energy Infrastructure (2025). The long-duration storage layer that lets variable clean generation firm flexible compute.

The Two-Ledger Data Center Bargain (2026). The local net-benefit framework behind community consent.

Three Competing Governance Equilibria (2026). Conditional acceleration as the coordination equilibrium this paper assumes.

The Federal-State AI Infrastructure Collision (2026). The FERC and RTO layer the connected-is-not-firm argument draws on.

AI Data Center Credit Risk (2026). The credit lens behind the investor exposures in Risk Mitigation.

The MindCast AI Data Center Record (2026). The corpus index for the full data-center workstream.

Sources

External references.

4–16 GW single frontier training run (peak power density): Wide-Area Power System Oscillations from Large-Scale AI Workloads — arXiv, 2025.

80–90% inference share of total AI energy (fleet energy): AI Data Centers Energy Consumption in 2024–2026 — TTMS, 2026.

PJM Board of Managers, direction on an interim resource-adequacy framework (July 27, 2026).

United States export controls on advanced computing and AI model weights, 15 CFR Part 740.

License Exception Artificial Intelligence Authorization and Supplement No. 5 to Part 740, current as of 2026.

Nicholas Institute, Duke University, Rethinking Load Growth (2025).

Lawrence Berkeley National Laboratory, 2024 Report on U.S. Data Center Energy Use, for the Department of Energy.

Working With MindCast

MindCast turns each forecast into a commissioned engagement. The offers key to the predictions.

MindCast turns the analysis into three commissioned engagements. Each one converts the paper’s mechanism into a decision a specific team can act on.

💼 Executives can commission a Fleet Firmness Map. The engagement sorts every site by whether its workload can move or must stay, then matches each anchored site to a firm-clean or capacity-backed supply plan and each flexible site to an energy-advantaged region. Operators leave knowing which megawatts to firm and which to relocate.

🏛️ Policymakers and RTOs can commission a Large-Load Service Design. The engagement drafts the tariff and interconnection terms that differentiate shortage treatment by capacity backing and enforceable flexibility. A state gains a way to attract data-center investment without shifting reliability cost onto existing ratepayers.

⚖️ Counsel can commission an Offshore-Placement Decision Map. The engagement checks each candidate workload against the current export-control structure and names which jurisdictions and workload classes qualify. A firm gains a defensible path to place mobile compute abroad without stumbling into a violation.

Reach MindCast at mcai@mindcast-ai.com. Visit our Corporate Site.

Executive Summary

Demand did not break in July. The financing and permission system began charging for the cost of meeting it. Four hyperscaler earnings reports in nine days showed the market paying for one thing: conversion evidence relative to incremental capital.

The backdrop is the June 2026 repricing, when markets cut AI infrastructure valuations on doubts that the buildout would pay. Firms that register their own economics re-rate, and the re-rating is the registration dividend. Firms that leave the market’s question unanswered reprice.

The July reactions split against every simple story. Microsoft rose about 15 percent and Amazon nearly 14 percent despite Amazon crossing into negative free cash flow. Alphabet fell roughly 7 percent despite a $514 billion cloud backlog, and Meta fell about 9 to 10 percent as free cash flow collapsed 91 percent.

The thesis follows directly. The market no longer asks whether demand exists. The market asks whether contracted demand can convert through six gates: customer performance → authorization → energization → utilization → revenue → free cash flow.

A hierarchy came with the question. Capital guidance without demand loses. Backlog without sufficient conversion coverage may still lose, and registered conversion relative to incremental capital earns the premium.

MindCast approaches the question differently from narrative research. AI Repricing Cycle 2026: Microsoft, Nvidia, Oracle, Meta and the Validation Tests Every AI Layer Now Faces (hereafter AI Repricing Cycle) published a crossing calendar and a scored prediction register before events resolved. Microsoft, Oracle, Amazon and the Escape from the AI Repricing Cycle (hereafter Escape) converted the diagnosis into registration-lag doctrine.

Every claim carries a falsifier, a checkpoint and a settlement source fixed in advance. A multi-party cognitive digital twin foresight simulation generates the probability bands, and cognitive digital twins (CDTs) model the actors inside it.

The paper proceeds in three movements. Sections I and II state the July evidence and score the prior register against it. Sections III through V develop the conversion mechanism and revise the crossing calendar, and Sections VI and VII follow the repricing into credit markets and statehouses.

Seven AIRC-IV predictions test the conversion layer. They cover backlog deceleration; the Meta and Alphabet free-cash-flow crossings; firm-specific credit differentiation; authorization disclosure; conversion reporting; and Meta’s external-compute strategy. Section IX states their checkpoints, falsifiers and settlement sources.

🏛️ Policymakers: Authorization is now a priced input to the buildout. Pause power converts into bargaining power denominated in developer financing terms.

💼 Executives: Conversion disclosure outperforms commitment disclosure. Publish metrics that sit deep in the conversion chain before the market demands them.

⚖️ Counsel: A published backlog is a representation surface. The conversion telemetry behind it is the exhibit the next securities complaint requests.

📊 Investors: Price the gates, not the backlog. Dispersion across issuers is the investable signal in equity and credit alike.

I. What July Registered

Four earnings reports in nine days supplied the cycle’s first full comparative test, with each firm exposing a different capital-conversion configuration.

Microsoft reported fiscal fourth-quarter revenue of $90.0 billion on July 29, up 18 percent. Conversion evidence carried the day. Azure growth accelerated to 43 percent and Azure crossed $100 billion in annual revenue within total Microsoft Cloud revenue above $214 billion. Commercial remaining performance obligations (RPO) rose 84 percent to $678 billion.

Adoption and efficiency registered beside the demand. Copilot paid seats expanded from the prior quarter’s more than 20 million to more than 30 million, with quarterly net additions more than doubling. Management quantified up to 40 percent better performance per watt from its own silicon and application-level GPU-cost reductions of 84 to 89 percent. Calendar-2026 capital expectations stayed economically unchanged at approximately $175 billion after lease reclassification, and the stock rose approximately 15 percent.

Amazon reported the next day and crossed on schedule. Revenue reached $200.6 billion, up 20 percent, while Amazon Web Services (AWS) grew 37 percent at a 39 percent operating margin. The AI and custom-chip businesses each passed a $25 billion annual run-rate.

Trailing-twelve-month free cash flow printed negative $7.6 billion, the exact crossing AI Repricing Cycle dated to this window. Management raised capital guidance to $220 billion and cited a $496 billion AWS backlog on the call, and the stock rose nearly 14 percent.

Alphabet reported July 22 and supplied the decisive counterexample. Google Cloud grew 82 percent to $24.8 billion at a 35.6 percent operating margin, and the cloud backlog reached $514 billion.

Free cash flow still turned negative for the first quarter in company history as capital spending doubled to $44.9 billion. Guidance rose to $195–205 billion in the second upward revision of the year, and shares fell roughly 7 percent. Registered backlog and measurable monetization did not offset a cash crossing paired with another capital increase.

Meta reported strong advertising and fell hardest. Revenue reached $60.8 billion with ads up 27 percent. Quarterly capital expenditure of $31.1 billion consumed roughly 98 percent of operating cash flow, and free cash flow fell 91 percent to $784 million.

Full-year guidance rose to $130–145 billion, and management publicly outlined an external-compute opportunity without attaching a quantitative metric. The stock fell about 9 to 10 percent.

The four reports isolate the variable. Microsoft and Amazon connected spending to adoption, margins and efficiency. Both re-rated.

Alphabet published the commitment metric without the cash pairing and repriced. Meta published the spending without comparable conversion evidence and repriced harder.

II. The First Scoring Cycle

Prior installments registered fourteen predictions with falsifiers, checkpoints and settlement sources fixed in advance. The table scores them against the July record. Statuses are Settled, Open, Trigger Review or Untriggered.

AIRC-II.4 settles true because Microsoft added qualifying adoption and capital-efficiency disclosures. AIRC-I.2 enters trigger review because reporting linked Meta’s decline to its free-cash-flow collapse and AI spending, with the review publishing alongside the December scoring note.

AIRC-I.6 remains open because the investment gains demonstrate conduit exposure but not the stress transmission required for settlement. All other entries retain their published checkpoints.

AIRC-II.4 settles true. AIRC-I.5 has accumulated strong supporting evidence, while the remaining entries await their registered triggers and checkpoints.

III. Conversion, Not Commitment, Earned the Premium

Escape staked AIRC-II.2 on a first mover reframing capital guidance as milestone-gated. No firm pulled the lever this cycle, and the entry stays open to its July 2027 checkpoint. What the quarter revealed is which registration the market pays for: firms earned the premium by connecting spending to conversion.

Two structural readings fall out of the rows. First, the market accepted a cash-flow crossing when conversion stood beside it and rejected a crossing paired with an unmatched capital increase. The crossing itself is not the corrective event; the crossing without conversion evidence is.

Second, commitment metrics rank below conversion metrics but above bare spending. Alphabet’s backlog cushioned a fall Meta took at full force.

One ratio organizes the market’s response to each new capital increase:

Marginal Conversion Coverage = Δ Expected Cash Conversion ÷ Δ AI Capital Commitment

Marginal Conversion Coverage measures whether newly registered conversion economics justify the latest increase in capital. Microsoft and Amazon registered stronger coverage in July. Alphabet registered exceptional cloud conversion but weaker consolidated coverage relative to another capital increase, and Meta registered the capital before publishing a quantitative external-compute conversion metric. The ratio remains a working construct until simulation tests its calibration.

Game theory supplies the payoff structure. Conditionality shrinks the strategy set and signals surrender inside the Nash capex trap AI Repricing Cycle mapped, while disclosure changes only the market’s information. Predictive Game Theory Meets the Era of AI — Operationalizing Fudenberg’s Research Agenda with Cognitive Digital Twins supplies the formal result: belief-updating moves weakly dominate strategy-shrinking moves in a mutating game, which is why no firm gated.

Behavioral economics supplies the salience rule. Markets punish the most visible unmatched number and reward the metric closest to cash. Predictive behavior emerges from the combination: the premium tracks how deep in the conversion chain the disclosed variable sits.

💼 Executives should read the table as a disclosure playbook. The next earnings cycle rewards gate-depth metrics, not larger commitment figures.

IV. Financeable Backlog: The Conversion Gates

Every escape instrument becomes the next cycle’s exposure, and the quarter’s instrument is no exception. Three cohort firms now carry published forward-commitment metrics measured in the hundreds of billions of dollars. Escape‘s metric-substitution finding warned that institutions stabilize around a measurable proxy after the market changes the outcome it wants measured, and named backlog and orders as the hazard class. Alphabet’s quarter shows the market already discounting the proxy.

The mechanism states cleanly. Registered backlog becomes valuable only to the extent that the six gates make it convertible. One working construct organizes the gates:

Financeable Backlogₜ = Contracted Demand × P(Customer Performance ∩ Authorization ∩ Energization ∩ Utilization by t) × Cash-Conversion Margin × Discount Factorₜ

The joint probability preserves dependence among the gates, and the discount factor prices the cost of waiting for them to clear. The time horizon distinguishes backlog expected to convert within two years from an equivalent nominal commitment requiring ten. The construct routes the CDTs and awaits simulation before any calibration attaches. Financeable Authorization Coverage in AI Data Center Credit Risk — Permitting, Curtailment, and the Cost of Capital(hereafter Data Center Credit Risk) measures the authorization gate at project level; Financeable Backlog is the firm-level analog running the full chain to cash.

Representation surface grows with each published figure. A $678 billion or $514 billion number is a statement every subsequent quarter can contradict, and the contradiction now carries the weight the market assigned the metric on the way up. Deceleration, cancellation or counterparty concentration each qualify as candidate corrective events. The discoverable-knowledge problem in The Duty to Foresee — AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care applies in full: whatever internal conversion telemetry sits behind a published backlog is the exhibit the next complaint requests.

Counterparty concentration sharpens the hazard. Oracle carries $638 billion of RPO with roughly $300 billion concentrated in one counterparty, against a BBB− rating. The diligence question updates accordingly: for every published backlog, ask what share one counterparty carries and which gates stand between the contract and the megawatts.

⚖️ Counsel should treat every published backlog as a litigable representation. 📊 Investors should demand the recognition schedule behind it.

Visit MindCast Corporate

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure, MindCast Foresight Prediction Simulations, Synthesizing Behavioral Economics + Game Theory.

Working With MindCast

MindCast runs two service lines on one method. Market foresight intelligence grades earnings disclosures, crossings and credit events against the conversion-gate map. Disclosure exposure intelligence maps the representation surfaces beneath them: backlog, guidance and conversion metrics, the seams where investor relations ends and securities exposure begins.

The foresight simulation predictions in this paper are the market line applied to the conversion layer now pricing across equity and credit. Every engagement below runs on the same methodology.

💼 CFO, treasury and investor-relations teams can commission a conversion-disclosure audit: the current disclosure set scored against the six gates and the gaps between it and what Microsoft and Amazon registered in July. The audit prices Marginal Conversion Coverage on the next capital increase and times issuance against the crossing calendar.

📊 Investors and lenders can commission a crossing-and-coverage screen across a named portfolio: where each holding sits on the crossing calendar, what Financeable Backlog implies by gate probability, and which issuers carry spread-differentiation exposure under AIRC-IV.4. The screen states what each firm’s next report must register to hold its multiple.

⚖️ Securities counsel can commission a representation-surface review: published backlog and conversion claims mapped against the telemetry discovery would reach, corrective-event scenarios keyed to the AIRC-IV.1 deceleration trigger, and disclosure posture ahead of the crossing quarter.

🏦 Credit analysts and rating teams can commission an issuance-differentiation screen: cohort debt mapped against the matched-benchmark spread test, an authorization-risk language audit keyed to AIRC-IV.5’s disclosure baseline, and the collateral assumptions behind GPU-backed financing channels.

Entries move only when the public record moves, and engagements buy application of the map rather than revision of it. Engagements run as CDT simulations with dated, falsifiable outputs.

V. The Crossing Calendar Accelerated

The crossing calendar dates each firm’s arrival at the point where AI capital spending pushes free cash flow negative. The quarter scored the calendar in both directions. One projection landed inside its stated window, one arrived early, and the full cohort now sits closer to the line than the published schedule assumed.

Oracle already crossed. Fiscal 2026 free cash flow ran negative $23.7 billion, the calendar’s anchor. Oracle now meets its registration test carrying an S&P downgrade to BBB− and a $45–50 billion raise plan, with Barrows v. Oracle, the securities action over its AI disclosures, in motion-to-dismiss briefing through December.

Amazon crossed on trailing free cash flow, on schedule. Trailing free cash flow printed negative $7.6 billion for the period ended June 30, the exact window AI Repricing Cycle published from Epoch AI’s filings-based model. The market’s response supported the conversion reading: the crossing arrived beside registered conversion and re-rated upward.

Alphabet crossed quarterly, ahead of schedule. The inherited projection read approximately Q1 2027, and Alphabet posted the first negative quarter in company history in July. Trailing free cash flow remains positive at $53.3 billion. AIRC-IV.3 tests whether the quarterly crossing becomes a trailing one, and Section VI carries the funding stack behind the spend.

Meta reached the threshold zone. Quarterly free cash flow of $784 million against $31.1 billion of capital expenditure puts the inherited crossing date two to three quarters early on straight-line arithmetic. AIRC-IV.2 registers the pull-forward. The compressed window also shortens the clock on converting the outlined external-compute opportunity into a registered metric.

Microsoft anchors the cohort. The cohort’s latest projected crossing belongs to the firm that just cleared its validation appointment with calendar-year capital economically unchanged. Microsoft did not cross, registered conversion anyway and re-rated.

The takeaway is structural. Crossings arrive early because capital guidance keeps rising faster than the prior models assumed, and every upward revision pulls the crossing date toward the present unless conversion improves fast enough to offset it.

VI. Credit Markets Reprice the Buildout

Credit spreads on hyperscaler debt activated this month as the deciding signal AI Repricing Cycle named. Data Center Credit Risk established on August 11 that lenders now weigh permitting readiness and community standing directly in credit assessment. The translation applies at cohort scale: authorization uncertainty becomes a credit spread when uncertainty changes the timing, probability or recoverability of the cash flows supporting capital.

The absorption numbers arrived first. AI hyperscaler debt issuance reached $220 billion through August 10. Technology spreads reached 89 basis points, nine wider than the overall investment-grade market, and Amazon’s $25 billion offering required roughly 120 basis points over Treasuries. Debt already funds roughly a third of hyperscaler capital expenditure and is heading toward 35 percent against $1.14 trillion of projected 2027 spending.

Alphabet’s funding stack tells the story from the issuer side. The company carries $98.2 billion of long-term debt after raising more than $85 billion over the prior year. The $84.75 billion equity package includes a potential $40 billion at-the-market program rather than cash fully received. Capital markets increasingly supplement retained earnings in financing the buildout; what changed is the scale and the price.

One cohort rating action has already landed. S&P downgraded Oracle to BBB− against $638 billion of RPO, negative free cash flow and a $43 billion debt year. Meta’s Hyperion vehicle raised approximately $29.5 billion in the largest non-merger high-grade bond sale on record, distributing financing outward while retaining contractual exposure.

NVIDIA extends the channel furthest. Six financial institutions signed memorandums of understanding to mobilize more than $500 billion of GPU-backed financing for buyers without the rating or cash to purchase silicon outright. NVIDIA’s five-year credit default swap spread roughly doubled since late May.

Two collateral assumptions sit underneath the channel: authorization holds and GPUs retain value over a debt tenor. Neither is settled, and Data Center Credit Risk carries the residual-value test as entry P16 of its own project-level prediction register.

Discrimination, not distress, is the signal worth reading. Recent investment-grade corporate offerings continue to clear, but at increasingly differentiated cost and one downgrade in.

The credit market is performing Section III’s conversion analysis with a longer memory. Volume rises and velocity falls. Dispersion widens, and only dispersion is investable. AIRC-IV.4 registers the event-form test at cohort level, and AIRC-II.7 stays untriggered because no qualifying equity drawdown has demanded credit confirmation.

📊 Investors should track issuance spreads by name, not by sector. The dispersion is where the repricing becomes tradable.

VII. Authorization Delay Accrues Against Borrowed Capital

States joined the repricing in the same three weeks credit did, and the coincidence is the mechanism. Texas paused a 474-gigawatt interconnection queue for audit on August 3. The Electric Reliability Council of Texas (ERCOT) targets completion by December, and the Public Utility Commission of Texas (PUCT) will seek expanded statutory authority when the legislature convenes in January.

Pennsylvania followed on August 18. Executive Order 2026-05 conditions state permits on binding Governor’s Responsible Infrastructure Development (GRID) commitments and prior local approval. The order removes data centers from fast-track permitting, bans state nondisclosure agreements with developers and ties tax exemptions to compliance.

Two of the largest hosting states repriced authorization in one month. Neither event settles an entry in the companion series’ bargaining-power register (MC-BP); both supply new evidence for the mechanism.

Authorization delay accrues against drawn money as Expected Delay Carry, and sufficiently irreversible committed capital lowers a sponsor’s ability to wait. Corporate capital carries interest regardless of whether the infrastructure program it supports clears authorization on schedule. Every month of authorization delay after capital is drawn carries financing cost, which converts the state’s pause power into bargaining power denominated in the developer’s own financing terms.

Authorization is also an early gate in Section IV’s conversion chain. A backlog whose megawatts cannot clear authorization is contracted demand with a conversion probability the market has begun pricing at the state line. The feedback loop closes the system:

Higher financing costs raise the price of delay, and a higher price of delay increases the developer’s willingness to concede. Concessions raise project cost and feed the next round of capital requirements. The loop continues until one of four conditions arrives: demand weakens; financing conditions ration marginal projects; conversion improves enough to restore coverage; or authorization supply expands. AIRC-IV.5 tests for the first documentary evidence of the coupling: authorization risk named in the corporate financing record.

🏛️ Policymakers hold a strengthening negotiating position and should price it deliberately. Concession terms set now become the template the next legislature inherits.

VIII. Open Predictions and Unsettled Tests

Four appointments structure the remainder of 2026. NVIDIA reports in late August and tests AIRC-I.4, the registered claim that NVIDIA reprices through customer financing limits rather than its own execution. Every major customer just raised or maintained enormous capital programs, so any drawdown on the print is correspondingly diagnostic.

Oracle reports its fiscal first quarter in September, the first reporting window that can score AIRC-II.5, the registered claim that Oracle’s financing materials will distinguish ownership from obligation by December 31. The Barrows briefing prices escalation lag in a courtroom over the same months.

Meta’s next two quarters test AIRC-IV.2 and the external-compute registration clock together. Two scoring documents publish by December 31: the AIRC-II.1 decomposition separating June’s disclosure-gap losses from timing effects, and the AIRC-I.2 review determining whether Meta’s decline qualifies as a corrective disclosure.

IX. Predictions Registry: Conversion Layer

AIRC-IV designates the fourth block of the AI Repricing Cycle registry and tests the conversion layer. AIRC-III remains assigned to the regulatory-repricing installment, and AIRC-IV contains six primary predictions and one conditional extension.

Cohort means Microsoft, Amazon, Alphabet and Meta. Oracle serves as an adjacent comparator and does not count toward cohort settlement thresholds. Probability bands publish from the simulation and appear beside each prediction.

📐 AIRC-IV.1: The registered backlog becomes the exposure. Following publication, the first cohort firm to report year-over-year growth in a consistently defined RPO or backlog metric at less than half the average of its prior four reported year-over-year growth rates produces a two-trading-day excess return against the Nasdaq-100 tracking fund (QQQ) worse than the median QQQ-excess return of the other three cohort firms over the same window. Metric redefinitions do not trigger the entry, and firms lacking four comparable prior observations are ineligible to trigger settlement.

Event-triggered through December 31, 2027. Falsifier: the first qualifying deceleration produces a QQQ-excess return at or better than the other-firm median. Settlement: earnings materials and market data.

📉 AIRC-IV.2: Meta’s crossing pulls forward. Meta reports negative trailing-twelve-month free cash flow in its Q4 2026 or Q1 2027 results. Checkpoint: Meta’s Q1 2027 earnings. Falsifier: trailing free cash flow remains positive through the Q1 2027 report. Settlement: Meta earnings releases.

🌀 AIRC-IV.3: Alphabet’s trailing crossing follows its quarterly one. Alphabet reports negative trailing-twelve-month free cash flow by its Q2 2027 results. Falsifier: trailing free cash flow remains positive through the Q2 2027 report. Settlement: Alphabet earnings releases.

🏦 AIRC-IV.4: Credit differentiation becomes firm-specific. Following publication, at least two cohort issuers price qualifying corporate debt with matched-benchmark-adjusted issuance spreads differing by at least 50 basis points by June 30, 2027. The comparison controls for currency, seniority and duration, and the adjusted spread is each issue’s spread minus the same-day spread of a named investment-grade technology benchmark at comparable duration. Qualifying issuance is corporate debt priced after publication and expressly tied by the issuer to AI infrastructure, with the benchmark and data source named in the settlement note.

Falsifier: no qualifying adjusted differential appears by the checkpoint. Settlement: offering documents, pricing supplements and benchmark curves. Concordance: the entry tests the dispersion read from Data Center Credit Risk at event level; the two settle independently with no shared credit.

🗺️ AIRC-IV.5: Authorization risk enters the cohort financing record. Following publication, at least one cohort issuer newly adds or materially expands an explicit permitting, interconnection or authorization-delay risk in a corporate debt offering document or rating publication by June 30, 2027. The disclosure must connect the risk to AI infrastructure returns, cash flow, capital requirements or credit quality. Generic regulatory language without that connection does not qualify, and settlement runs against a baseline audit of cohort disclosures dated through publication.

Falsifier: no qualifying disclosure appears by the checkpoint. Settlement: corporate prospectuses and rating publications. Concordance: cohort corporate instruments only; project-finance documents settle the project-level register of Data Center Credit Risk (P1 through P3), and no single observable settles both registers.

🔄 AIRC-IV.6: Conversion disclosure arrives. Following publication, at least two cohort firms newly disclose a backlog-conversion or consumption metric beyond the standard twelve-month RPO split by December 31, 2027. Qualifying metrics include recognition schedules, consumption rates, cancellation or deferral rates and counterparty concentration. Qualitative statements that demand remains strong do not qualify.

Falsifier: fewer than two firms publish qualifying metrics by the checkpoint. Settlement: earnings materials and SEC filings.

Extended tier. ☁️ AIRC-IV.7: Meta’s external compute registers before the crossing. Conditional on AIRC-IV.2’s crossing arriving, Meta publishes a quantitative external-compute metric before or concurrently with the earnings release that first reports the crossing. Qualifying metrics include revenue, contracted capacity or named-customer commitments, and the band reports as P(IV.7 given IV.2) so readers do not mistake a conditional band for an unconditional forecast.

Falsifier: the crossing arrives with no qualifying metric published. Settlement: Meta earnings materials and company disclosures.

Dated checkpoints: NVIDIA earnings in late August 2026; the Oracle fiscal Q1 report in September 2026; the Barrows v. Oracle briefing through December 2026; the AIRC-II.1 decomposition and AIRC-I.2 trigger review by December 31, 2026; Meta’s Q4 2026 and Q1 2027 reports; Texas audit completion targeted for December 2026; the Texas and Washington legislative sessions convening January 2027.

X. Conclusion

AI Repricing Cycle ended on the twin law: the crossing date you never watched governs your multiple. Escape added the agency clause: the lag you never measured governs your drawdown. The scored quarter appends the conversion clause, and the hierarchy came with it.

Capital guidance without demand loses. Backlog without sufficient conversion coverage may still lose. Registered conversion relative to incremental capital earns the premium.

Backlog is no longer the endpoint but an upstream claim on the gates: customer performance → permission → power → utilization → cash conversion. Financeable Backlog prices whether those gates clear and when. Marginal Conversion Coverage prices whether the resulting economics justify the next dollar of capital.

The next repricing begins when one conversion gate fails, or when capital markets start charging materially more for the probability that it will. Equity, credit and authorization now operate as one linked system, and the AIRC-IV register tests where the linkage becomes observable next.

Engagements

MindCast runs two service lines on one method. Capital-markets foresight intelligence grades earnings disclosures, crossings and credit events against the conversion-gate map. Infrastructure exposure intelligence maps the gate surfaces beneath them: authorization, energization and utilization. The seams sit where contracted demand must become deliverable megawatts.

The foresight simulation predictions in this paper are the capital-markets line applied to the conversion layer now pricing across equity, credit and statehouses. Every engagement below runs on the same methodology.

💼 Hyperscaler and developer finance teams can commission a conversion-disclosure audit: the current disclosure set scored against the six gates and the gaps between it and what Microsoft and Amazon registered in July. The audit prices Marginal Conversion Coverage on the next capital increase and sequences new metrics to the earnings calendar.

📊 Investors and lenders can commission a crossing-and-coverage screen across a named portfolio: where each holding sits on the crossing calendar, what Financeable Backlog implies by gate probability, and which issuers carry spread-differentiation exposure under AIRC-IV.4. The screen states what each firm’s next report must register to hold its multiple.

⚖️ Counsel can commission a representation-surface review: published backlog and conversion claims mapped against the telemetry discovery would reach, corrective-event scenarios keyed to the AIRC-IV.1 deceleration trigger, and disclosure posture ahead of the crossing quarter.

🏛️ State offices and legislative staff can commission an authorization-leverage assessment: where the Texas and Pennsylvania templates migrate next, what concessions are worth when priced against developer financing cost, and a session-ready sequence for January 2027.

Entries move only when the public record moves, and engagements buy application of the map rather than revision of it. Engagements run as CDT simulations with dated, falsifiable outputs.

Appendix — MindCast Foundations

AI Repricing Cycle 2026: Microsoft, Nvidia, Oracle, Meta and the Validation Tests Every AI Layer Now Faces. Supplies the validation regime and the crossing calendar Section V scores, plus the AIRC-I register scored in Section II.

Microsoft, Oracle, Amazon and the Escape from the AI Repricing Cycle. Supplies the registration-lag doctrine Section I refines to conversion depth and the metric-substitution finding Section IV extends to published backlogs.

Agent Governance Equilibrium. Supplies the resilience instrumentation behind the recovery readings: firms re-registered economics faster than markets could harden June’s gaps into class periods.

The Dual Nash-Stigler Equilibrium Architecture. Names the two exits from the capex trap and explains why disclosure moves escape the trap’s penalty without changing the strategy set.

The Duty to Foresee — AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care. Supplies the discoverable-knowledge problem Section IV applies to published backlogs.

Predictive Game Theory Meets the Era of AI — Operationalizing Fudenberg’s Research Agenda with Cognitive Digital Twins. Supplies the mutating-game result under which Section III’s refinement is legible: belief-updating moves weakly dominate strategy-shrinking moves.

AI Data Center Credit Risk — Permitting, Curtailment, and the Cost of Capital. Supplies Section VI’s translation from authorization uncertainty to credit spread, plus Expected Delay Carry and the dispersion read AIRC-IV.4 tests. Project-layer wagers stay in its P register.

The AI Data Center Authorization Bargaining Power ratings. Supplies the authorization-market structure Section VII couples to financing cost, read here as new mechanism evidence without settlement credit.

Appendix — Source Note

Primary records reviewed run through August 21, 2026. They include the four firms’ earnings materials and SEC filings. Amazon’s $496 billion backlog is management’s earnings-call figure.

Alphabet’s financing description follows its second-quarter call and June 2 securities filing, which distinguishes completed underwritten offerings from the potential at-the-market component.

Pennsylvania’s description follows Executive Order 2026-05 and the Governor’s implementation summary.

Credit-market figures use contemporaneous Reuters reporting. Oracle, Meta Hyperion, NVIDIA financing and residual-value figures retain the source record published in Data Center Credit Risk.

Appendix — Acronyms

AIRC: AI Repricing Cycle, the series’ running prediction registry.

AWS: Amazon Web Services, Amazon’s cloud infrastructure division.

ERCOT: Electric Reliability Council of Texas, the grid operator conducting the interconnection audit.

GRID: Governor’s Responsible Infrastructure Development, Pennsylvania’s data center standards made binding August 18.

MC-BP: the bargaining-power prediction register of the companion authorization series.

CDT: cognitive digital twin, the actor model inside the multi-party foresight simulation that generates the series’ probability bands.

PUCT: Public Utility Commission of Texas, the regulator partnered on the audit and seeking expanded 2027 authority.

QQQ: the Nasdaq-100 tracking fund, the market benchmark in AIRC-IV.1.

RPO: Remaining Performance Obligations, contracted revenue not yet recognized.

Related works: The Authorization Market: Standardized Bargaining, Rationed Power, and the Competition to Build America's AI Infrastructure · The Data Center Authorization Price: A 50-State Baseline · AI Data Center Authorization Bargaining Power — Ratings for Hyperscalers, Neoclouds, Developers, and Capital · The 50-State Regulatory Atlas · The Model AI Infrastructure Authorization Code · AI Data Center Credit Risk — Permitting, Curtailment, and the Cost of Capital · The MindCast AI Data Center Record

See the MindCast Data Center Regulatory Economics Live-Fire Mission

I. Executive Summary

Authorization — the institutional right and practical ability to build, energize, and operate hyperscale infrastructure — is repricing on timelines capital planning rarely anticipates. Cle Elum, Washington moved from project announcement to enacted moratorium in three days. New York imposed the nation’s first statewide data-center pause by executive order. Oregon, Arizona, and Illinois froze tax-incentive eligibility mid-year while their land-use moratorium bills stalled. Capital deployed into a jurisdiction that reprices mid-project absorbs delay carry and terms it never underwrote.

MindCast AI built the Authorization Transition Forecaster to answer the decision question before markets price it: where does authorization risk reprice next, through which legal instrument, at which approval gate, and inside what time window? Users upload the instrument to any web-enabled AI system; the instrument researches the live public record and returns a county-level early-warning map, a ranked forecast register, and — for every forecast — a model-derived probability band, a dated window, a named falsifier, and a public settlement source that will prove the forecast right or wrong.

The simulation behind this paper has already run, and its register is frozen. Three headline Foresight Simulation Predictions set the stakes:

• The county wave keeps rolling — at least 20 additional U.S. counties enact moratoriums or application pauses by January 31, 2027, on top of the 54 already enacted nationwide. 75–85%

• Pauses mature into pricing rules, not bans — a majority of local moratoriums expiring by June 30, 2027 are replaced by explicit siting, infrastructure, and resource rules rather than reverting to the old regime; the emergency pause is a bridge, not an endpoint. 70–82%

• Silence breaks before strictness — in pre-registered matched jurisdiction pairs, the authorization-silent jurisdiction experiences material intervention before its strict-framework peer, reversing the site-selection instinct that permissive means safe. 65–75%

The full register appears in Section VIII.

Each reader class holds a different edge in these pages.

Developers and hyperscalers read the map to time site commitments and sequence entitlement before a jurisdiction reprices mid-project.

Investors and lenders read it to underwrite transition risk — which jurisdictions in a portfolio sit one docketed study session away from a freeze.

Utilities and commissions see controversy forming before the rate case turns political, and see where blocked land-use pressure migrates to their gates next.

State and local lawmakers learn the difference between “low regulation” and authorization silence — why silence is what breaks, why the pause is a bridge to explicit terms, and what the jurisdictions that wrote rules early got for it.

Counsel reads the vesting boundary: where litigation concentrates, and which projects the instruments cannot reach.

The sections below spell out the vision, the archetypes that motivate it, the simulation behind it, the intended usage for eleven stakeholder classes, the mitigation architecture, and the full prediction register.

II. The Vision

Authorization is the priced input of AI infrastructure. MindCast established the framing in The Authorization Market, which defined permission to build and operate as an economic good carrying a price, and measured that price across every state in The Data Center Authorization Price: A 50-State Baseline and the standing 50-State Regulatory Atlas. Power can be contracted, land assembled, and capital raised, but counties, cities, utilities, commissions, and legislatures grant the right to build — and across 2026 they repriced that right on timelines few site models anticipated.

Application pauses arrived days after project announcements. Statewide instruments arrived by executive order where legislatures had failed. Tariffs and incentive programs moved while site plans sat in review.

Markets treat these interventions as political surprises. MindCast treats them as observable state transitions inside a coupled institutional system — and state transitions can be forecast. A jurisdiction moves through five states: silence, salience, institutional disequilibrium, material intervention, and a new authorization equilibrium. Each stage emits evidence before the next arrives: a docketed study session, a staff-directed ordinance draft, a newly opened utility proceeding, a neighboring county’s vote entering the same media market.

Moratorium trackers count what already happened. The Forecaster models the jurisdictions that have not yet acted, reads their emissions, and names the instrument, gate, and window before the vote. Forward sight, not retrospective counting, is the product.

III. Five Archetypes, Not One Event Class

Five 2026 interventions show why “moratorium” is the wrong unit of analysis. Each case stopped or repriced data-center development, and no two used the same instrument, authority, or stop point — which means no two carry the same durability, falsifier, or settlement source.

Collapsing these five into one "moratorium" category mispredicts all three things a forecast must get right: how long the intervention lasts, what evidence would falsify it, and which public record settles it. The companion instrument therefore types every intervention (T1–T6) before banding it, and treats the type as part of the forecast rather than a footnote to it.

Archetype sources, dated — one short paragraph per archetype.

Texas. Hill County adopted the first Texas county pause in early 2026. A developer’s $100 million federal lawsuit forced its rescission in early June, and the county substituted a project-review framework — the authority lesson settling in real time. San Marcos became the first Texas city to ban data centers outright through zoning.

New York. Executive Order No. 62 — one year, facilities of 50 megawatts or more — took effect July 14, 2026, one month after the legislature passed the Responsible Data Center Development Act, which awaits the Governor’s action and an expected veto.

Washington. Cle Elum’s council enacted its emergency moratorium on July 31, 2026, three days after Blue Fern Development’s announcement, under the state’s interim-ordinance authority. Whether the pause reaches the triggering project turns on a 2002 development agreement, with a public hearing set for September 8. Seattle’s one-year pause passed in June, followed by Spokane and Federal Way.

Oregon. H.B. 4084’s Enterprise Zones freeze took effect June 5, 2026, running until 90 days after the 2027 session adjourns. Hillsboro enacted the state’s first local pause in late July and Salem followed in early August. Governor Kotek cancelled a state land sale for the Salem project on July 30, then endorsed local moratoriums on August 7 while stating the state lacks statewide authority. Four legislators proposed a three-year statewide moratorium on August 4 for the 2027 session.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

Working With MindCast

MindCast runs two service lines on one method. Authorization intelligence grades jurisdictions and instruments against the fifty-state baseline. Geopolitical exposure intelligence maps the classification, entity-eligibility, and screening constraints that sit above them. The forecast in this paper is the authorization line applied to the transition side of the market — where the rules reprice next — and every engagement below runs on the same frozen methodology.

Developers and hyperscalers can commission a portfolio placement — named sites scored against the authorization-state model, precursor exposure at each jurisdiction, vesting posture per project, and a sequenced entitlement plan keyed to the windows the register predicts. Investors and lenders can commission a transition-risk screen across a named portfolio: which holdings sit within reach of a precursor threshold, which instrument class each jurisdiction is most likely to produce, and what the repricing does to delay carry and exit assumptions.

Utilities and commissions can commission a gate-migration assessment — where blocked land-use pressure arrives at their dockets next, keyed to tariff, interconnection, and water proceedings and to the federal filing cycle now in motion. States, counties, and commissions drafting 2027 frameworks can commission a transition drafting review — converting a temporary pause into explicit authorization terms, keyed to host-community provisions, durable employment categories, and the clauses the coming sessions will contest.

Forecast entries move only when the public record moves; engagements buy application of the map, never revision of it. Engagements run as Cognitive Digital Twin simulations with dated, falsifiable outputs, and MindCast grades its record in public.

IV. The Foresight Simulation

Vision becomes method through simulation, and the simulation is where MindCast’s architecture does its work. The method runs on published theoretical foundations — The Dynamic Predictive Game Theory Collection for contests where the rules change during play, and Field-Geometry Reasoning for how constraint fields close paths as commitments become irreversible.

The architecture has run before. The same MP CDT FS engine executed the nine-twin simulation behind AI Data Center Credit Risk and the live-fire validations against the Super Bowl and the World Cup final.

The Forecaster begins with Cognitive Digital Twins of seven core actors: the governing body, the planning institution, the utility, the state regulator, the developer, the organized residents, and the neighboring jurisdiction. Additional institutional actors — a water authority, a court, a governor, an environmental regulator, a transmission owner — enter when the authorization pathway requires them.

Simulation runs each jurisdiction forward under five scenarios — the base case, a project-acceleration shock, a contagion shock, a utility or resource shock, and a developer concession. Each run identifies the actor whose strategy changes first, the trigger that moves it, the most likely intervention and its legal gate, the timing, the strongest counterforce, and the falsifier. Concession testing matters most for practitioners: the simulation shows whether infrastructure funding, phased load, or community benefits reduce intervention probability or merely change the intervention’s form.

The MindCast AI Data Center Record shows the discipline in operation: twelve months of dated claims graded against the public record, including three published corrections.

V. What Each Stakeholder Does with the Forecast

Eleven stakeholder classes face the same repricing risk from different directions, and each can run the identical instrument. Every call-out below names the distinct decision the instrument improves.

Data-center developers and site selectors — the site-commitment decision. A county with no data-center regulation is not low-risk; it is authorization-silent, and silence under project pressure is among the strongest precursors of an abrupt pause. The Forecaster informs the choice between committing now, accelerating entitlement before optionality closes, opening concession negotiations, or paying a higher but known authorization price elsewhere.

Infrastructure investors, private equity, and lenders — the underwriting decision. The Forecaster converts regulatory change into a portfolio variable: exposure by risk tier, exposure through land-use gates versus utility gates, and the share of committed capital still subject to government optionality. AI Data Center Credit Risk supplies the companion instruments — Financeable Authorization Coverage and the Capital Irreversibility Ratio — that score which projects convert an authorization shock into a credit event.

Foreign investors, sovereign wealth funds, and foreign-affiliated developers — the disclosure-and-jurisdiction decision. Foreign capital carries authorization exposure in its sharpest form, because ownership identity is itself an intervention trigger. A pension fund holding project debt is anonymous to a county hearing; a sovereign fund or foreign strategic investor is not, and the disclosure remedy that lowers a domestic sponsor’s capital price can raise a foreign sponsor’s authorization price by handing organized opposition its most effective argument. National-security review stacks on top of every local gate the Forecaster maps, and neither layer appears in conventional country-risk analysis. Foreign Ownership and Control carries MindCast’s full treatment, including its own prediction register.

Jurisdiction selection is therefore a narrower problem for foreign-affiliated capital than for anyone else at the table, and the Forecaster narrows it deliberately: which states are advancing foreign-ownership disclosure or restriction statutes, which localities convert foreign identity into political salience fastest, and which authorization gates — land-use, utility, or state legislative — a foreign-linked project will actually face. Sequencing matters as much as selection. Disclosure timed before the salience forms reads as transparency; disclosure forced after opposition organizes reads as concealment, and the two carry different authorization prices.

Hyperscalers and capacity buyers — the capacity-reliability decision. Two campuses identical on power price, land cost, and incentives can differ sharply in interruption risk. The Forecaster ranks a pipeline by which jurisdictions are likely to hold their authorization terms through the deployment window.

Utilities — the tariff-design and engagement-timing decision. Service territories where load growth, ratepayer salience, and a weak local framework converge become public controversies on a forecastable schedule. Tariff structures, infrastructure agreements, and stakeholder engagement all cost less ahead of the controversy than inside the rate case.

State legislators and agencies — the framework-timing decision. The instrument diagnoses where project pressure is outrunning authorization architecture and which localities will improvise pauses absent a statewide framework. The 2026 record constrains the realistic menu: executive instruments succeeded where legislative ones stalled, and fiscal instruments passed where land-use bills failed.

County and municipal governments — the rule-writing decision. A deep dive on your own jurisdiction, run while policy optionality is still open, shows whether current code can distinguish a hyperscale campus from ordinary industrial use. Governments that write rules before the first controversial application are far less likely to reach for the emergency pause their neighbors improvised.

Economic-development organizations — the positioning decision. Permissive is not competitive when permissiveness signals an unpriced repricing ahead. Marketing regulatory certainty — explicit, bounded, stable terms — targets the capital that stays.

Law firms and regulatory counsel — the exposure-mapping decision. Diligence concentrates where the Forecaster maps pressure: contested intervention authority, vesting postures that determine litigation exposure, and community pressure blocked at the zoning gate migrating toward water and power proceedings.

Public utility commissions and RTO-facing stakeholders — the docket-anticipation decision. Authorization conflict migrates from land-use gates into tariff, interconnection, and cost-allocation proceedings, and vested projects that local pressure cannot stop through zoning arrive at the commission next. The Forecaster flags the migration before the docket opens.

Researchers, journalists, and policy analysts — the hypothesis-testing decision. Seven causal hypotheses — regulatory lag, strictness-as-stability, physical constraint, two contagion mechanisms, salience-times-optionality, and fiscal asymmetry — publish as falsifiable claims with named evidence standards, and every national run grades them.

Symmetry across the table is deliberate. Developers and regulators can interrogate the same evidence from opposite sides of the authorization table. Foresight serves the table.

VI. Mitigation: Changing the Pathway, the Timing, and the Price

Identifying authorization-transition risk without addressing what changes it would leave developers and capital with a map and no steering wheel. Mitigation deserves its own architecture, and the architecture rests on one governing principle: mitigation does not eliminate authorization risk — it changes the intervention pathway, the timing, or the price. The strongest mitigation often does not prevent government action at all; it converts a threatened moratorium into a conditional-use regime, an infrastructure agreement, a phased-load schedule, or a bounded framework.

MindCast’s own simulation makes the point directly: register entry MC-ATX-S4 predicts that developer concessions change the instrument more often than they restore the prior permissive regime. The developer’s realistic objective is rarely to make regulation disappear. It is to move a jurisdiction from an unpredictable emergency intervention into a priced, bankable authorization framework.

Three strategic classes organize every mitigation below. Prevent transition reduces the probability that a jurisdiction moves from silence or salience into material intervention. Shape transition accepts that intervention is likely and steers it toward a more predictable instrument. Preserve exit maintains alternative sites, phasing, and capital flexibility in case the authorization price becomes unacceptable. Sophisticated sponsors run all three simultaneously, because the classes hedge one another.

Reduce regulatory lag before the project becomes salient (prevent). Sponsors should determine whether a jurisdiction holds a mature data-center authorization framework before filing, and where rules are incomplete, negotiate explicit standards proactively rather than relying on silence or by-right ambiguity. Moving a jurisdiction toward an explicit authorization equilibrium before controversy peaks reduces the likelihood that officials reach for an emergency pause simply to create a decision framework.

Sequence entitlement before salience peaks (prevent and preserve exit). Vesting stage governs exposure: sponsors should know exactly when rights attach, which approvals remain discretionary, and which gates — power, water, transmission, environmental — stay open after land-use rights close. The objective is not merely filing faster; it is avoiding peak controversy while government still holds maximum optionality, because that combination is where the simulation finds interventions concentrate.

Phase load rather than forcing an all-or-nothing grid decision (shape). A single 600 MW request creates a different regulatory game from a staged build in tranches. Phased energization reduces transmission urgency, tariff exposure, stranded-asset concern, and ratepayer salience — the exact variables that convert a private development matter into a commission controversy.

Internalize infrastructure costs before government imposes them (shape). Developer-funded substations, transmission upgrades, water infrastructure, and minimum-take commitments change the utility’s and regulator’s payoff structure directly. Expect instrument substitution rather than deregulation — the concession typically becomes a condition, and a priced condition is the bankable outcome.

Target the politically legible externality, not a generic benefits package (prevent and shape). The binding constraint differs by jurisdiction — water in one, residential proximity or transmission corridors in another, electricity prices or farmland conversion elsewhere. Mitigation aimed at the variable actually driving local salience outperforms broad community-benefit spending aimed at none of them.

Make fiscal benefits legible and durable to the constituency absorbing the cost (prevent). Intervention risk rises when perceived burdens outrun perceived local benefits, and the perception is what votes. Long-duration property-tax commitments, host-community agreements, and transparent employment and revenue projections shift the ratio — but only when the benefits demonstrably land on the residents and ratepayers carrying the externalities.

Preserve geographic optionality until the critical gates close (preserve exit). Authorization-transition risk deserves the same treatment as power availability: maintain alternative sites until entitlement and energization gates actually close. A jurisdiction that looks cheap because it is silent can become more expensive than an explicitly strict alternative after repricing — the matched-pair prediction MC-ATX-P11 exists to test exactly that reversal.

Manage ownership disclosure before it becomes adversarial (prevent). For foreign-backed and sovereign-linked sponsors, early ownership transparency keeps a local land-use dispute from unexpectedly stacking foreign-ownership statutes, federal review, and political scrutiny onto the project. Disclosure timed before salience forms reads as transparency; disclosure forced after opposition organizes reads as concealment, and the two carry different authorization prices.

Mitigation, run well, is authorization-price management rather than opposition management. Sponsors who reduce lag, sequence entitlement, phase load, internalize costs, and keep exit options open do not escape the transition the Forecaster maps — they choose which instrument meets them, on which timeline, at which price. Regulators reading the same list should notice the symmetry: every mitigation above is also a template for what a jurisdiction can ask of a project before the emergency pause becomes its only option.

VII. Bargaining Against Transition Risk

Transition risk is not identical across firms, because firms carry different bargaining capabilities into the same repricing. MindCast’s companion ratings, Who Bargains Best with the States: AI Data Center Authorization Bargaining Power (the buy-side ratings of the AI Infrastructure Authorization Series) rate the field — hyperscalers, neoclouds, developers, and capital providers — across seven capabilities: collateral capacity, clean-portfolio depth, governance capital, demand credibility, optionality, federal alignment, and speed. The Forecaster measures the probability that the bargaining environment changes; the ratings measure the firm’s position when it does. Six connections show where the two instruments meet.

Optionality becomes transition insurance — a firm holding several viable jurisdictions credibly walks when authorization reprices, while a single-site developer absorbs whatever instrument arrives.

Collateral converts intervention into negotiable conditions — balance-sheet strength turns infrastructure contributions, minimum-take requirements, and guarantees into terms a firm can accept, where the same conditions would be fatal to a thinner sponsor. Clean-portfolio depth expands substitution options — a builder that clears the strict clean-statute states can migrate toward explicit-framework jurisdictions rather than staying trapped in permissive-looking but unstable ones.

Governance capital reduces political salience — terms-first conduct lowers the probability that a project’s externalities become the trigger for emergency intervention at all. Demand credibility changes regulator payoffs — a state or utility negotiates around credible long-duration load and pauses on speculative queue positions. Speed cuts both ways — it carries a firm through an open authorization window, and the ratings’ sharper observation is that aggressive speed manufactures the freezes this paper forecasts.

Together the two instruments describe a single underwriting object: a firm’s authorization resilience is a function of its exposure to regulatory transition and its capacity to absorb, negotiate, or escape that transition. MindCast leaves the relationship conceptual rather than formalized — how the dimensions interact is an empirical question the graded registers of both papers will answer before any composite deserves a formula.

Readers pricing a specific portfolio should read the two papers as halves of one transaction: the Forecaster names where the rules of the bargain are most likely to change, and the ratings name who can bargain through the change.

VIII. Foresight Simulation Predictions

The MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation has executed for this paper, and its output is a register of Foresight Simulation Predictions (FSPs) — dated claims carrying model-derived confidence bands, explicit falsifiers, and named public settlement sources, frozen August 16, 2026 and published under the prefix MC-ATX(MindCast Authorization Transition).

The register’s composite claim is a sequence of institutional adaptation: local pause wave → state instrument substitution → gate migration → framework formation → pricing equilibrium. Twelve primary FSPs test the links in that sequence. Bands are foresight judgments, not calibrated frequencies, and every entry settles under the typology, precursor scoring, and inventory protocol specified in the companion instrument, against the record maintained in the National Authorization Intervention Inventory.

• MC-ATX-P1. At least 20 additional counties or county-equivalents enact a new moratorium or application pause by January 31, 2027. 75–85%.

• MC-ATX-P2. At least 3 additional Oregon localities enact pauses before October 31, 2026. 70–80%.

• MC-ATX-P3. The Cle Elum moratorium fails to reach the vested Blue Fern project — concession or adverse ruling by December 31, 2026. 60–70%.

• MC-ATX-P4. Zero states enact a statewide land-use moratorium by statute through the 2027 adjournments, under a frozen instrument definition. 65–75%.

• MC-ATX-P5. At least 1 additional state acts through the executive corridor — order, audit-conditioned pause, or administrative halt with operative effect — by December 31, 2027. 55–65%.

• MC-ATX-P6. At least 2 additional states enact tax-incentive pauses or restrictions in 2027 sessions. 65–75%.

• MC-ATX-P7. At least 2 projects materially insulated at the land-use gate face new contested utility, water, or environmental proceedings by June 30, 2027. 60–70%.

• MC-ATX-P8. At least 1 state enacts data-center-specific foreign-ownership disclosure or restriction by the 2027 adjournments. 55–65%.

• MC-ATX-P9. Among jurisdictions reaching the Forecaster’s maximum administrative-precursor score, at least 70% adopt a material intervention within 90 days. 70–80%.

• MC-ATX-P10. At least 2 states convert study bodies or moratorium windows into enacted statewide frameworks — or preemption statutes bounding local authority — by December 31, 2027. 55–65%.

• MC-ATX-P11. In at least 3 pre-registered matched pairs, the authorization-silent jurisdiction experiences material intervention before its strict-framework peer. 65–75%.

• MC-ATX-P12. A majority of local pauses expiring by June 30, 2027 are replaced by explicit pricing rules rather than reverting to the prior regime. 70–82%.

Eight secondary FSPs supply the diagnostic layer beneath the primary set, most in the interpretive class and reported separately from event probabilities:

• MC-ATX-S1. State instruments concentrate outside framework states. 70–80%

• MC-ATX-S2. Framework states produce bounded, sunset-dated local instruments. 70–80%

• MC-ATX-S3. Learning contagion outruns developer displacement. 65–75%

• MC-ATX-S4. Concessions change the instrument rather than restore the regime. 65–75%

• MC-ATX-S5. A strict jurisdiction publicly captures displaced development. 55–65%

• MC-ATX-S6. Litigation concentrates where entitlements are advanced and capital is sunk. 70–80%

• MC-ATX-S7. Utility-side actions grow their share of the intervention mix by at least 10 percentage points. 60–70%

• MC-ATX-S8. Opposition predicts intervention only where government retains authorization optionality. 65–75%

One structural read closes the set: MC-ATX-SR1 — the wave settles into pricing, not prohibition (70–80%), with the pause-replacement ratio of MC-ATX-P12 as its observable. Grading MC-ATX-P12 grades the thesis.

Three entries hold QUALIFIED status until their pre-registration conditions execute — the framework-state cohort list, the matched pairs, and the national Top 25 — and the register reports those dependencies openly. Readers should judge the Forecaster by how this register grades.

IX. Foundational Works

Six published MindCast works carry the load underneath the Forecaster, and each supplies a proposition the instrument operationalizes. No claim in this vision statement requires reading them; readers who want the underlying studies will find each one self-contained.

The Authorization Market defines permission to build as a scarce economic good carrying a price, with predictable terms outperforming nominal permissiveness because capital can underwrite them. The Data Center Authorization Price: A 50-State Baseline measures that price in every state on one instrument, and the 50-State Regulatory Atlas maintains the standing jurisdictional record the Forecaster’s screens run against.

The Model AI Infrastructure Authorization Code supplies the gate hierarchy that separates material authorization from routine administration. AI Data Center Credit Risk translates authorization uncertainty into credit spread and carries its own sixteen-entry prediction register. The MindCast AI Data Center Record grades twelve months of dated claims against the public record and demonstrates the settlement discipline the Forecaster’s register will follow.

Theoretical foundations sit one layer deeper: Chicago School Accelerated for authorization as bargaining rather than administrative process, The Dual Nash-Stigler Equilibrium Architecture for where contests settle and when search stops, Field-Geometry Reasoning for constraint fields and closing paths, and The Dynamic Predictive Game Theory Collectionfor contests whose rules change during play. Frameworks publish in full; the runtime architecture applying them does not.

Related works: The MindCast AI Data Center Record · The Authorization Market: Standardized Bargaining, Rationed Power, and the Competition to Build America’s AI Infrastructure · AI Data Center Credit Risk — Permitting, Curtailment, and the Cost of Capital · Three Competing Governance Equilibria for AI Infrastructure · The Two-Ledger Data Center Bargain · The Federal-State AI Infrastructure Collision

Critical references: Data Center Authorization Transition Forecaster · National Authorization Intervention Inventory · The Model AI Infrastructure Authorization Code · The Data Center Authorization Market: A 50-State Regulatory Atlas

See the MindCast Data Center Regulatory Economics Live-Fire Mission

Executive Summary

Permission to build an AI data center has become a scarce, priced good. Roughly $725 billion in annual hyperscaler capital is chasing a fixed pool of authorization slots — grid headroom, tariff capacity, water allocations, community tolerance — and the states that control those slots have started rationing them: auditing speculative demand, scoring bidders, and pausing whole markets while they write terms. Under rationing, the question that decides where capital lands is no longer which firm bids most. It is which firm the state will say yes to, and on what terms.

The ratings in this paper answer that question for the whole field. Every major buyer of authorization — Microsoft, Google, Amazon, Meta, OpenAI, Oracle, Anthropic, xAI, the neocloud tier, the developers who manufacture slots, and the capital behind all of them — is rated on the seven capabilities states actually price: collateral, clean-firm power, conduct, demand credibility, optionality, federal alignment, and speed. The headline finding runs against market instinct: the bargaining ranking does not match the market-cap ranking. Microsoft and Google lead because they carry collateral, clean supply, and governance simultaneously; the announce-first, gas-forward half of the field is racing a shrinking map, and one firm’s conduct record is now the exhibit opposition groups cite in other firms’ hearings.

Twelve Foresight Simulation Predictions freeze the analysis into dated, falsifiable claims. Three set the stakes:

• Slot scarcity is measurable and binding — in every rolling twelve-month window through mid-2028, no more than 60 campuses of 250 MW or larger receive binding authorization nationwide, while disclosed demand runs at least three times the cleared count. 70–80%

• No hyperscaler builds at home — no principal headquartered in California or Washington breaks ground on a 250 MW campus in its own state through mid-2028; the states that host the buyers’ headquarters will not host their loads. 75–85%

• Governance becomes the sales pitch — two or more principals publicly market authorization and governance capability as competitive differentiation, converting conduct from a compliance cost into a marketed asset. 80–89%

Each prediction carries a deadline, a falsifier, and a public settlement source, and the full register appears in Section XI. Readers who want the state side of the same bargain — what the fifty states charge for authorization — will find it in the companion baseline linked below; this paper stands alone without it.

Each stakeholder class holds a different edge in these pages. Operators read the ratings to know which states price their strengths and which price what they lack. Investors and lenders read them to underwrite the stack behind announced capacity — the dependency graph no state statute yet prices. States, counties, and commissions read the same table in reverse: bidder diligence for the RFPs and statutes the 2027 sessions will produce. And the firms rated lowest read it for the only thing that moves a score — the record.

The current publication is the buy-side half of the series’ paired baseline. The AI Data Center Authorization Price — A 50-State Baseline mapped the sell side — what fifty states charge for authorization — with its full evidentiary layer in the The Data Center Authorization Market: A 50-State Regulatory Atlas and its scoring instrument in The Model AI Infrastructure Authorization Code. The papers run on the same frozen baseline and the same instrument-change ledger.

Roughly $725 billion in 2026 hyperscaler capex — the four-company estimate frozen at T₀; later reporting runs closer to $745 billion — is chasing a fixed and shrinking pool of authorization slots: grid headroom, tariff capacity, water allocations, community tolerance. The Authorization Price baseline rated the state side of the bargain; this paper rates the firm side.

Four tiers compete for the same slots by different means: principals whose demand drives the race, neoclouds who build fastest with the thinnest governance, developers who manufacture and hold slots, and capital whose balance sheets determine who can post the collateral that financial-assurance regimes now use to ration headroom.

The question throughout: which firms are best positioned to bargain with states — and the answer is two-sided, because a bargaining position is what the state will pay to have you and what you can extract for showing up.

Disclosure: the roster includes Anthropic, whose Claude models MindCast uses in its research workflow. Anthropic is scored on identical criteria to every other firm, and the reader can audit the scoring against the sourced record like any other row.

I. 🧭 What “Bargaining Position” Means Here

The AI Data Center Authorization Price established the pricing structure: the pause threat sets the floor on conditions, the federal bypass sets the ceiling on price, and buyer competition determines where the price clears inside the band. A firm’s bargaining position is therefore not a single number but a bundle of seven capabilities, each mapped to a mechanism in the main paper:

Collateral capacity. Financial-assurance regimes ($1.5M/MW Virginia, $50k/MW non-refundable in the Texas draft) ration headroom by balance sheet. Firms that post capital effortlessly clear the rationing screen; firms that need backstops inherit their guarantor’s constraints.

Clean-portfolio depth. Binding clean-statute states (the companion’s Form 1 tier — WA, CA, OR, NY, MN, MI and ten more) are structurally the high-price, high-certainty tier, and only firms with 24/7-carbon-free procurement capability clear their bar. Gas-dependent builders are locked out of sixteen states before negotiations begin.

Governance capital (the Conduct column). The terms-first versus announce-first axis. TLSM-11 (one operator’s controversy cited against unrelated companies, 85–92%) makes community tolerance a shared commons; a firm’s conduct record is priced into every rival’s sites, and its own.

Demand credibility. The state’s P(honored): will the load, jobs, and revenue actually materialize? Self-funded demand scores high; circularly financed or speculative demand scores low, and states are learning to tell the difference — AEP’s halved forecast after tariff approval was speculative demand exiting the queue.

Optionality. Outside options are bargaining power. A firm running a fifty-site portfolio credibly walks; a firm locked into one metro cannot. The extreme version inverts the auction entirely (§6).

Federal alignment. The accelerate tempo lives federally; alignment with it buys land access, permitting compression, and political cover — and creates exposure if the administration or the policy changes.

Speed capability. Time-to-compute is the race’s clock. Speed is an asset in accelerate-tier states and a liability everywhere terms matter, because the fastest builders are the ones that manufacture freezes.

One equation organizes the seven, and the series umbrella publishes it as this paper’s governing form: Slot Odds ∝ Capability Bundle × (1 − Conduct Discount). Capabilities add — a thinner collateral position can be partially offset by a deeper clean book or wider optionality. Conduct multiplies — the discount applies to the whole bundle at once, updates on the firm’s record in every jurisdiction simultaneously, and no capability buys it back except performance over time. The multiplicative term is why ⚠️ exists as a category in the master table rather than as a low score: a sufficiently damaged conduct record discounts everything the firm brings to every table it sits at.

Regional market rules are converting the same bundle from negotiating posture into service qualification. Under frameworks like PJM’s proposed Interim Resource Adequacy Service, accredited capacity, enforceable curtailment, and verified flexibility purchase firmness itself, not just goodwill — a capacity-short load that cleared every state gate still stands first in line for shortage curtailment. The seven capabilities below therefore price twice: once at the bargaining table, and again at the regional layer no state statute reaches.

The composite Position rating is a tier judgment, not an arithmetic sum; the legend above the master table defines every column in plain terms.

Two Propositions Under Test

The series umbrella pre-registered two propositions this guide carries, both stated as model outputs with definitions and falsifiers rather than established facts.

The slot-clearing proposition (70–80%). The national market clears only several dozen new large-campus authorizations annually, against disclosed demand several times larger. Definitions, per the umbrella’s contract: minimum campus size is 250 MW; the unit of authorization is a campus holding a binding instrument — an executed large-load service agreement, approved tariff enrollment, or completed interconnection agreement — never an announcement; geographic coverage is the fifty US states; duplicate removal counts each campus once across announcements, re-announcements, and phase splits; oversubscription is estimated as deduplicated disclosed demand (announcements plus interconnection-queue requests) divided by cleared slots. Registered as MC-FG-9. [Operational thresholds bracketed pending freeze: “several dozen” proposed at ≤60 per rolling twelve-month window; “several times larger” proposed at ≥3×.]

The home-state conduct discount (65–78%). A principal’s home jurisdiction applies the largest conduct discount, because local institutions possess the deepest firm-specific record — every unmet commitment, every utility fight, every disclosure gap sits in the local file, and hometown visibility raises the symbolic payoff of opposition. Registered as MC-FG-10, with MC-FG-8’s groundbreaking test as its behavioral corollary: the discount is the mechanism, the absence of home-state campuses is the observable it predicts. §8 locates the paradox in the firm ratings; §9 develops its mechanism.

II. 🗺️ Master Table — The Field at T₀

How to read the ratings. Each firm is rated on seven capabilities on an underlying five-star scale, displayed as a single color star:

🟢 strong (4–5 stars) · 🟠 middling (3 stars) · 🟡 weak (2 stars) · 🔴 weakest in the field (1 star)

The final column, Position, is the overall judgment: 🥇 structurally advantaged · 🥈 capable with constraints · 🥉 dependent or exposed · ⚠️ conduct record actively damaging its own and others’ bargaining position. Ratings are interpretive classifications at 70–88% confidence, resting on the sourced record through T₀ — not arithmetic scores.

Tier 1 — Principals

Tier 2 — Neoclouds

Tier 3 — Developers (Slot Manufacturers)

Tier 4 — Capital

III. 📜 Principal Profiles — The Sourced Record

Microsoft — 🥇 the full-stack incumbent. Calendar-2026 capex near $190 billion, capacity-constrained through at least year-end by its own CFO’s account, with the deepest clean-firm portfolio in the field, and the firm publishes the numbers: 40 GW of new renewable supply contracted since 2020 with 19 GW online, 100% annual matching of global electricity consumption achieved in 2025, 29.8 million metric tons of contracted carbon removal, a carbon-negative-by-2030 commitment, the Crane Clean Energy Center (Three Mile Island) delivering from 2027, geothermal in New Zealand, and the first fusion PPA on record with Helion. Worth stating precisely, because the distinction decides Form 1 access: the achieved and published position is annual matching plus contracted firm nuclear, not hourly matching — the 24/7 architecture is Google’s origination and Microsoft’s trajectory.

Collateral capacity is effectively unlimited and further leveraged through the AI Infrastructure Partnership with BlackRock/GIP and MGX. The governance record is strong with one visible dent: Microsoft lobbied alongside Amazon against Washington’s HB 2515, winning the session and guaranteeing a harder 2027 bill — a tactical win purchased at strategic cost in exactly the state where its clean portfolio should dominate. Against that dent sits the field’s most developed community-relations infrastructure: Microsoft Local publishes a standing page for roughly nineteen US jurisdictions and two dozen countries, each carrying the same five Community-First Infrastructure Initiative commitments (January 13, 2026) — pay our own way so datacenters do not raise local electricity prices, minimize and over-replenish water, hire locally, add to the tax base, and fund local AI training and nonprofits — alongside energy and water explainers, operations fact sheets, per-project construction updates, a Datacenter Academy pipeline, and a named local contact channel. Best-fit territory: Form 1 states and high-collateral regimes where rivals thin out. The Washington deficit is only half reputational: CETA compliance headroom and Climate Commitment Act allowance pricing distribute an Eastside load’s cost across every Westside ratepayer, so the legislators pricing Microsoft’s consent represent payers rather than hosts — a discount lobbying cannot repair and conduct cannot fully offset (§9).

Google — 🥇 and playing a second game nobody else plays. Capex $175–185 billion with cloud revenue up 63%, the original 24/7-CFE mover with the broadest advanced-clean pipeline (geothermal, advanced nuclear), and the same Form 1 access as Microsoft. The differentiator is the credit-enhancer role: Google backstops Fluidstack’s lease obligations, took stakes in TeraWulf and Cipher, and thereby underwrites the neocloud tier building Anthropic’s capacity — converting balance sheet into influence over slots it never has to permit itself. No other principal has financialized the authorization race this way. Exposure: the same backstops import neocloud conduct risk into Google’s TLSM-11 ledger.

Amazon — 🥈 the biggest spender with the most utility scar tissue. The largest 2026 program at roughly $200 billion, plus the $11 billion, 1,200-acre Project Rainier campus in Indiana purpose-built for Anthropic — delivered operational while rivals planned, which is genuine speed-with-terms capability. Clean portfolio substantial but shallower than the two leaders; the governance ledger carries the HB 2515 lobbying and a longer history of utility cost-allocation fights, and investor pressure for water and power disclosure lands on Amazon first among the three clouds. Best fit: bargain-enacted states with written tariffs, where its scale prices well and its disclosure gaps price less. Amazon shares Microsoft’s Washington apportionment geometry exactly, which makes the two firms the guide’s cleanest natural experiment: whatever the 2027 successor charges them differently is attributable to conduct, since the structure is held constant (§9).

Meta — 🥈 maximum force, gas-forward. Capex guidance $115–135 billion against multi-hundred-billion multi-year infrastructure signaling, with the largest single self-build campuses in the country and full demand integration — no cloud customers, so P(honored) on load is near-certain.

The constraint is fuel: the flagship program leans on gas generation, which forfeits the sixteen-state Form 1 tier before bargaining begins and concentrates Meta in accelerate and bargain-enacted gas states, where its willingness to fund community benefits at scale buys local consent the fuel mix otherwise costs.

Structural read: the strongest bargainer in the tier of states with the weakest terms — which is leverage today and exposure as the tiers converge.

OpenAI — 🥈 the strongest procedure, the softest foundation. Stargate stands at seven US sites — Abilene operating at ~0.3 GW with four of eight buildings live, Shackelford and Milam in Texas, Doña Ana in New Mexico, Port Washington in Wisconsin, the $16 billion Saline Township campus groundbroken in Michigan on June 1, and Lordstown’s Foxconn JV in Ohio — toward 9-plus GW by 2029.

The RFP that produced them drew over 300 applications from 30 states, the reverse auction analyzed in §6, and federal alignment through the Stargate banner is the strongest in the field. The soft points: capital is partnered rather than owned (Oracle, SoftBank, NVIDIA’s $100 billion), the Abilene expansion reversal shows plans move after groundbreaking, Capitol protests in Texas and water disputes at Doña Ana are accumulating on the commons ledger, and demand credibility rests on a $1.4 trillion ambition running far ahead of revenue.

Position summary: peak leverage now, fragile if the credibility gap widens — states that priced OpenAI’s promises at face value are carrying the P(honored) risk the Two-Ledger model prices at ~2× when it breaks.

Oracle — 🥈 the arms dealer who moved into the fort. Top-five capex with the $300 billion OpenAI agreement at its core, building and leasing Stargate capacity largely developed by others (Vantage, Crusoe), with internal component costs running $30–40 billion above plan by its own CEO’s estimate. Federal alignment high, clean portfolio middling, and the balance-sheet strain of financing someone else’s buildout is the field’s most-watched credit story. Oracle bargains well wherever Stargate’s political umbrella extends and weakly outside it.

Anthropic — 🥈 governance-differentiated, scale-constrained, structurally honest. The $50 billion US program with Fluidstack — Texas and New York first, sites live through 2026, the New York site likely TeraWulf’s hydro-powered Lake Mariner — sits atop the field’s most distinctive multi-cloud position: AWS Project Rainier scaling toward a million Trainium chips, a Google TPU commitment past a gigawatt, and a $30 billion Azure arrangement.

Hydro-backed sites give better Form 1 access than the capex number suggests. Two weaknesses: capital is backstopped rather than owned (Google’s guarantees behind Fluidstack and TeraWulf), and the announced 800 permanent jobs against $50 billion states the thin-benefit-line problem more plainly than any competitor — which is a governance asset with communities that have been burned by inflated jobs promises, and a bargaining handicap with legislatures still running the jobs/MW denominator. Best fit: Form 1 and bargain-enacted states that price enforceability over payroll.

xAI — ⚠️ the negative pole, and the field’s biggest externality. The record, per the litigation file: Colossus 1 in Memphis ran up to 35 unpermitted gas turbines from June 2024, removed them only under notice of intent to sue, and permitted 15; Colossus 2’s power plant in Southaven, Mississippi — sited across the state line where portable-equipment registration outruns Tennessee’s federally delegated Title V program, which plaintiffs call forum shopping — ran 27 publicly acknowledged temporary turbines that regulator correspondence later put at 59, with a fleet near 420–495 MW, potential NOx emissions that would make it the largest industrial source in a metro already failing smog standards, landing on predominantly Black communities.

The NAACP, SELC, and Earthjustice filed the Clean Air Act suit April 14, 2026; Mississippi issued a 41-turbine permanent permit three weeks after a single public hearing; a $7 million sound wall failed to quiet neighbors; the federal government has moved to dismiss.

Bargaining consequence: xAI has speed and one relationship, and has spent everything else — its conduct is now the citation opposing counsel and county boards use against every operator in the field (TLSM-11’s mechanism running at maximum), its Form 1 access is zero, and its optionality is gone because the sunk gigawatt anchors it to the one metro where its name is a rallying cry. The reported SpaceX ownership 🔶 adds consolidation without adding governance.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

Working With MindCast

MindCast runs two service lines on one method. Authorization intelligence grades jurisdictions and instruments against the fifty-state baseline. Geopolitical exposure intelligence maps the classification, entity-eligibility, and screening constraints that sit above them. The ratings in this paper are the authorization line applied to the buy side of the market, and every engagement below runs on the same frozen methodology.

Operators and principals can commission a position briefing — the firm’s seven-capability score decomposed against the sourced record, gap-to-leader per axis, tempo-tier exposure across the fifty states, and the record changes that would move each score. Investors and lenders can commission a counterparty assessment separating feasibility from fragility across a named position: the dependency graph behind announced capacity, collateral and backstop structures, and the probability that commitments are honored on the terms disclosed.

States, counties, and commissions running RFPs or drafting 2027 statutes can commission bidder diligence or scoring design — conduct records, demand credibility, and clean-firm supply converted into criteria an instrument can score, keyed to the provisions the coming sessions will contest. Neoclouds and developers can commission a qualification review naming the enforceable instruments, backstop structures, and drafting positions that raise a rating, because scores move only when the record moves.

Advisory sells record changes, never score changes, and runs one principal per contested arena per bargaining cycle. Engagements run as Cognitive Digital Twin simulations with dated, falsifiable outputs, and MindCast grades its record in public.

IV. 🧩 The Lower Tiers — Who Actually Holds the Slots

Neoclouds are the fastest tier and the most dependent: Crusoe built the Abilene flagship and owns the proven fast-build template; CoreWeave carries Stargate workstreams while building a software stack to escape the commodity trap; Fluidstack vaulted from GPU broker to Anthropic’s $50 billion partner on the strength of Google’s credit enhancement.

The TeraWulf/Cipher conversion class turns stranded crypto power into AI slots, which is the cheapest authorization arbitrage on the map — the permits, interconnection, and often hydro power already exist.

The tier’s structural position: it manufactures speed and consumes commons, concentrates in unwritten-terms jurisdictions per the companion analysis, and increasingly survives on principal backstops — meaning its bargaining position is borrowed, and the lender is a hyperscaler.

Developers are the quiet winners of the rationing regime. Vantage holds two of seven Stargate sites; entitled land with queue vintage is exactly the asset class the stringency ratchet appreciates, and every ratchet click raises the price principals pay to buy position rather than build it.

The capital tier’s function is narrower than its press: converting balance sheets into postable collateral (AIP), demand guarantees (NVIDIA), and patient ownership of the developer tier (Blackstone, Brookfield). One entry deserves its own line: Google-as-credit-enhancer is the cycle’s novel instrument — a principal underwriting the tier that builds for its competitor-partner, capturing optionality on both sides of the Anthropic relationship.

V. 🔁 The Reverse Auction — When States Bid for Firms

OpenAI’s RFP drew over 300 applications from 30 states, and sixteen states publicly signaled interest in hosting campuses. Read against the companion paper, the event is structurally significant: the competition layer established buyer competition as the fourth force pushing the authorization price up, and the RFP shows the force running in reverse at sufficient project scale.

A gigawatt-class campus with a $16 billion price tag is large relative to a state’s economy in a way a 100 MW facility is not, and at that ratio the state becomes the bidder — Michigan’s groundbreaking, New Mexico’s approvals over water objections, and Wisconsin’s Lighthouse all cleared processes that smaller projects in the same states would not have cleared on the same terms.

The rule this implies, stated as a hypothesis at T₀: bargaining power flips to the firm when project scale crosses a threshold relative to the host economy, and flips back to the state as the state’s remaining headroom depletes. Both movements are visible in the record — the RFP flip at announcement, and the Washington/Seattle counter-case where 369 MW of quiet demand against a constrained clean grid produced not a bid but a ban.

The synthesis with the companion’s ratchet: mega-projects buy their way past current terms and then become the incumbency that generates the next, stricter terms. OpenAI is simultaneously the biggest beneficiary of the flip and the biggest future contributor to the ratchet. Interpretive confidence in the threshold mechanism: 75–85%.

VI. 🎯 Firm × Tempo Fit — Who Wins Where

Bargaining position is not one national score — it is a fit between a firm's capability bundle and the tempo of the state it faces. A state's tempo tier, from the Authorization Price taxonomy, sets which capabilities the bargain actually prices: bargain-enacted states price collateral and contract tolerance, accelerate states price raw speed, pause states price queue vintage and demand credibility, contested states price pre-coalition instruments, and the clean-statute tier prices the power portfolio before conversation begins. The table below crosses the field against those tiers, naming who each tempo structurally favors and who it excludes.

Two structural reads fall out of the cross. First, only Microsoft and Google appear on the favored side of every tempo — the payoff of carrying collateral, clean supply, and conduct simultaneously rather than maximizing one. Second, the exclusion column converges: announce-first, gas-forward, thinly backstopped firms lose across every tempo for different stated reasons, which is one weakness expressing itself four ways. One tier is deliberately absent from the table — the sixteen-state clean-statute axis, where fit is hardening fastest and access is becoming a passport rather than a preference. Section VII gives it the full treatment.

VII. 🌿 The Clean-Energy Axis — Access, Conversion, and the Compounding Passport

Clean-energy bargaining position decomposes into three capabilities the field distributes unevenly, and the Clean column in §2 compresses what this section unpacks.

Access. The sixteen Form 1 states price hourly-matched, firm, additional clean supply — not annual renewable certificates. Google built the 24/7 carbon-free energy framework that effectively defines the standard those states are converging toward, with the broadest firm-clean pipeline in the field (geothermal, advanced nuclear, storage-paired renewables); its supply is bargaining-grade in more jurisdictions with less translation than anyone else’s.

Microsoft’s contracted book is arguably deeper in firm power — the Three Mile Island restart is the single most valuable clean asset any firm holds — and clears the same bar by a different route, though its published position is annual matching plus firm nuclear rather than delivered hourly matching, which narrows the gap to Amazon by one notch and widens Google’s lead by the same amount (interpretive, 70–80%; the two firms disclose on different bases, so the comparison rests on what each publishes rather than on a common metric). Amazon is the largest renewables buyer by volume but predominantly on annual matching, which prices completely in bargain-enacted states and incompletely in Form 1: access to roughly ten of the sixteen.

Anthropic punches above its capex per megawatt because the hydro-backed Lake Mariner position is exactly the asset class Form 1 states want, though the position is held through partners. Meta’s renewables purchases are enormous while its flagship program runs on gas, forfeiting the tier outright; OpenAI’s Abilene runs partly on onsite gas and inherits whatever its developers procure; xAI’s clean position is negative — the turbine record is the anti-passport, citable in every clean-state proceeding.

Conversion. Access clears the bar; conversion extracts value at it. The Washington mechanism from the companion paper is the template: in clean-constrained states the operator who finances new compliant generation is funding the state’s own mandate, which converts stringency into partnership and trades megawatts the state needs anyway for siting terms, speed, and goodwill.

Google and Microsoft both run the play; Google adds a wrinkle nobody else has — the TeraWulf and Cipher positions give it exposure to hydro-backed, already-permitted capacity, clean slots it never had to permit. Microsoft’s nuclear-anchored version is slower but stickier once landed.

Microsoft’s Community-First AI Infrastructure initiative deserves careful grading, because its commitments are more specific than a pledge normally is. Microsoft’s own explainer defines paying its way as paying utility rates high enough to cover its electricity costs and its share of the infrastructure costs to generate and deliver that electricity to its sites — which is the Model Code’s TF-2 no-cross-subsidy rule and TF-3 full infrastructure cost recovery, adopted voluntarily and stated in the enacted template’s own language.

Disclosure follows the same pattern. The same page publishes location-specific FY25 electricity and water data through an environmental data fact sheet and maintains per-jurisdiction operations fact sheets across twelve US states — TR-2 annual operations reporting and TR-3 public aggregates arriving without a statute. Add CE-1 clean share and a CE-2 additionality claim for the PPA program, and the voluntary book covers six leaf codes across three families, while leaving collateral, minimum take, and curtailment untouched — and while every one of the six remains unenforceable.

Enforceability is where the grade lands. Under the Code’s own instrument, a voluntary pledge scores its Consequence component at or near zero whatever its text promises, so the initiative is pre-compliance rather than price. Pre-compliance is nonetheless strategic behavior: publishing the template lowers the drafting cost for any state that decides to make it mandatory — the drafting-seat play executed on the firm’s own letterhead.

Sequence supports the strategic read. Community-First published January 13, 2026, seven weeks before the White House Ratepayer Protection Pledge of March 4; the firm-side instrument preceded the federal one (inference on directional influence rather than coincidence: 70–80%).

The dent in Microsoft’s conversion position is behavioral rather than portfolio: lobbying HB 2515 to death spent goodwill in precisely the state its clean book should dominate, weakening its claim to the drafting seat on the 2027 successor (interpretive confidence: 75–85%).

Compounding. The forward-looking point, stated as the axis’s structural claim: as the accelerate tier converts — and the Authorization Price baseline registers that conversion as Prediction 10 under Register v2.0 — clean-portfolio depth stops being a niche advantage in sixteen states and becomes the field-wide passport. Google’s and Microsoft’s lead compounds on the same clock that shrinks the map the gas-forward half of the field is racing.

Axis ranking at T₀: Google, Microsoft close behind, Amazon, Anthropic per-MW, then a gap, then Meta and OpenAI, then xAI. Interpretive confidence in the ordering: 78–86%. Registered as MC-BP-10 below: the clean-for-terms trade should surface as an explicit, public state partnership within the window.

VIII. 🧭 Firm × State — The Easiest and Hardest Bargains

Crossing the seven-axis scores against the companion’s state map (tempo, price certainty, friction, Form 1 membership, tribal exposure, local dispersion) produces each firm’s bargaining terrain. “Easiest” means the state’s binding constraints match the firm’s strongest axes; “hardest” means the state prices exactly what the firm lacks. Classifications are interpretive inferences at 70–85% per cell, resting on the T₀ record.

Tier notes. Neoclouds inherit their principals’ terrain where backstopped (Fluidstack maps to Anthropic’s, Crusoe to OpenAI’s) and default to the accelerate tier where not; the conversion class (TeraWulf/Cipher) is terrain-independent — its slots are grandfathered, which is the whole product. Developers hold position in whatever state their land bank sits, and the ratchet appreciates it there. Capital is terrain-agnostic by construction.

Post-T₀ note on the Texas cells. Every “easiest” Texas cell above — OpenAI’s three Stargate sites, Oracle’s HQ-adjacent position, Meta’s scale play — was written against the T₀ record. On August 3, Texas moved to an audit-gated ERCOT interconnection pause, which splits those cells by queue vintage rather than repricing them uniformly; the Post-T₀ Addendum carries the firm-by-firm read.

The home-state paradox — the finding this exercise surfaces. Every principal’s hardest bargaining terrain includes its home state: Washington for Microsoft and Amazon, California for Google, Meta, OpenAI, Anthropic, and xAI’s origin.

The mechanism is overdetermined — home states are Form 1 members with high friction, the firm’s visibility and symbolic value to opponents peak there, and hometown backlash reprices the whole portfolio (Seattle’s 369 MW episode fired at Microsoft’s and Amazon’s front door without either firm being named among the four applicants).

A second mechanism operates alongside record depth and behaves differently: externality geography. Home states are where a firm’s costs distribute most widely and its benefits concentrate most narrowly, so the largest number of legislators pricing its consent represent payers rather than hosts. Record depth is a conduct problem a firm can work off; apportionment is a fixed property of the jurisdiction. §9 develops the mechanism with Washington as the worked example, and carries the two register entries it generates.

The consequence is the buildout’s defining geographic irony: the AI industry’s physical footprint is being placed almost entirely in states where its firms do not live, which exports the externalities, imports the political exposure, and hands the hosting states durable leverage the home states declined to monetize. Registered as MC-FG-8. Interpretive confidence in the paradox’s mechanism: 80–88%.

IX. 🗺️ The Externality Geography of Authorization — Washington as the Worked Example

Costs and benefits do not share a boundary at a data center, and the mismatch decides who writes the terms. Construction spending, operating jobs, and property tax revenue land inside one county. Ratepayer exposure, clean-energy compliance burden, and carbon-allowance pressure spread across the entire state. Legislators therefore price authorization on behalf of constituents who pay without hosting, and the statute reflects that constituency rather than the host community’s.

Washington runs the mechanism at full strength, which is why it serves as the worked example — and why the section closes on whether the argument travels. Three claims follow: cost geography determines sponsorship geography, apportionment converts a diffuse cost into a legislative majority, and accountability asymmetry predicts the resulting statute’s specific shape.

Three transmission mechanisms. Physical spillover is the weakest of the channels carrying an Eastern Washington load onto a Westside utility bill, and the argument does not depend on it — statutory design carries the cost far more reliably than the grid does.

Three statutory channels carry an Eastern Washington load onto a Puget Sound utility bill — none of which requires physical spillover across the Cascades.

The Clean Energy Transformation Act (RCW 19.405) puts every Washington utility on one statewide clock, so a gigawatt-scale load in Grant County consumes non-emitting supply a Westside utility can no longer count toward its own 2030 and 2045 obligations; scarcity attaches to compliance headroom, which is a statewide pool drawn down by any load anywhere in it.

The Climate Commitment Act prices covered emissions through a statewide auction, so load growth that raises marginal emissions raises the clearing price for the gas utility serving a Seattle household that will never see a data center. And load above a utility’s federal preference allocation moves procurement to market, setting Mid-Columbia and transmission costs for parties who hosted nothing and approved nothing.

Compliance capacity is the most analytically important channel and the least discussed, because it is invisible to any framework treating environmental cost as a physical externality (interpretive confidence that compliance-capacity scarcity operates as a distinct, material channel in clean-energy-standard states: 72–84%).

Apportionment inverts the public-choice prediction. Data center siting presents textbook concentrated-benefit, diffuse-cost geometry. Concentrated benefit against diffuse cost ordinarily favors the concentrated interest, because beneficiaries organize and dispersed losers do not. Most of Washington’s population lives west of the Cascades, so the diffuse cost falls on a majority that is already organized — into legislative majorities in both chambers. Dispersed losers who happen to constitute a majority of districts need not solve a collective action problem; they need only vote their existing caucus. Sponsorship of data center legislation concentrating among Puget Sound Democrats while development concentrates in Grant, Douglas, Chelan, and Yakima counties is therefore not urban sentiment intervening in someone else’s county — the cost already arrived in the sponsors’ districts through the three channels above, and legislators of a state legislate for the state (interpretive confidence in the apportionment-inversion reading: 70–82%).

The accountability asymmetry, and what it predicts. Ordinary land use legislation carries a political restraint — a sponsor who wants a project sited in a district must eventually answer to that district — and the externality geometry removes it entirely. A Puget Sound legislator faces no electoral consequence from a Quincy or Moses Lake constituency. Ratepayer protection delivers benefit to that legislator’s voters directly. Host-county revenue sharing delivers benefit to voters the legislator does not represent, at a cost to project viability and therefore to the ratepayer protection that motivated the bill. Washington’s 2027 successor should accordingly run strong on cost causation, minimum take, collateral, exit protection, and disclosure, and comparatively weak on mandatory community benefit agreements, host payments, and dedicated local revenue shares — registered as MC-FG-11 below and graded against the Model Code’s PB-2, PB-3, and PB-8 leaf codes.

What separating the mechanisms buys the firm. Record depth predicts a discount that grows with the length and visibility of a firm’s local history, and conduct can repair it. Externality geography predicts a discount that grows with the ratio of statewide cost distribution to host-county benefit concentration, appears even for a firm with a short local record and a clean file, and no amount of conduct repairs it — apportionment is a fixed property of the jurisdiction, leaving three responses: price it, concede on it, or build elsewhere. A principal’s home state is where the ratio peaks, because home-state operations are large enough to move statewide compliance and rate variables while employment and revenue still land in specific counties. Interpretive confidence that both mechanisms operate and neither alone accounts for the home-state pattern: 65–78%.

Washington as the controlled comparison. Two principals headquarter in King County, both operate substantial Eastern Washington capacity, and both bargain with the same legislature under the same apportionment geometry — while their conduct records, disclosure postures, and clean-portfolio depths differ materially. Structure holds fixed; firm variables move. No other state supplies the configuration: California hosts headquarters without comparable in-state hyperscale load, Virginia hosts the load without hosting either principal. Visibly different treatment of Microsoft and Amazon by the same legislature under the same cost geometry is therefore attributable to conduct rather than structure, which is the proposition this guide advances — and Washington is its strongest available test.

Does the argument travel? A mechanism that works in one state is a description rather than a finding, and three jurisdictions test portability. Northern Virginia legislators represent both the Loudoun and Prince William host communities and a substantial share of affected ratepayers, so cost and benefit overlap more than in Washington — predicting stronger host-benefit provisions and weaker sponsorship-geography effects. Ohio’s load concentrates around Columbus with cost distributed across AEP Ohio’s territory and statewide through PUCO ratemaking: partial overlap. Georgia Power serves nearly the whole state as a vertically integrated monopoly, collapsing cost distribution into a single ratemaking forum. Host-benefit provision strength should correlate positively with host-jurisdiction/cost-jurisdiction overlap — registered as MC-FG-12 (interpretive, 62–75%, constrained by the small comparable set and by confounds from legislative professionalism and session length).

Three limits. Apportionment does not exhaust the explanation — party control, utility structure, labor coalition strength, and the presence of a clean-energy standard all bear on sponsorship patterns, and externality geography is one variable among several rather than the dominant one. Sponsorship geography here is observed, not tested: the Washington pattern rests on the E2SHB 2515 and SB 6171 sponsor and cosponsor records, and a rigorous version would code sponsorship geography against host-county location across every state with introduced data center legislation in 2025 and 2026 — mechanical work this guide has not done, which is why no national version of the claim enters the register. And the mechanism predicts drafting incentives rather than outcomes: legislators also respond to caucus discipline, Governor Ferguson’s priorities, tribal consultation obligations under EO 25-10, and organized labor, so a package the incentive favors can still fail or be amended for reasons the mechanism does not reach.

The specimen in the firm’s own information architecture. Microsoft’s Washington community page carries the URL slug /quincy/ and opens on Quincy, Malaga, and East Wenatchee — Grant, Chelan, and Douglas counties, with TechSpark Washington headquartered in the same corridor. Every Washington asset in the firm’s public community infrastructure sits east of the Cascades. No comparable page exists for the Puget Sound districts whose ratepayers absorb CETA compliance headroom and Climate Commitment Act allowance pricing, because there is no facility there to anchor one. Community relations follow the benefit; the statute follows the cost; the two never meet. A firm can staff, fund, and publish its way into standing with the counties that host it, and none of it reaches the legislators who write the terms — which is the externality geometry made visible in an org chart rather than a bill.

Which firms are equipped to bargain against it. Apportionment sets a price no firm can talk down, which turns the question from persuasion to capability — and the capability it rewards is not one the seven columns score directly. Bargaining against a payer majority requires assembling the host side of the divide into a visible counterweight: Grant County PUD load commitments, Eastside legislators with a revenue interest, building trades with a project calendar, chambers of commerce willing to host the corridor conversation. Incumbency supplies the raw material — twenty years in the Wenatchee valley is a standing coalition, not a talking point — and a newcomer with a clean file and a strong balance sheet cannot buy it inside one session. Microsoft and Amazon therefore hold an asset in Washington that partially offsets the same state’s discount against them, while a first-time entrant faces the discount with nothing on the host side of the ledger. The same explainer names the venue outright: Microsoft describes itself as participating in utility regulatory proceedings alongside consumer advocate groups to keep infrastructure planning current for all customers. A principal publicly seating itself beside the ratepayer advocates in commission dockets is the drafting-seat strategy stated in its own words, executed in the forum where the authorization price is actually set, and it is a posture no newcomer can adopt credibly on arrival. Microsoft Local is the industrialized version of the asset — roughly nineteen US jurisdiction pages carrying identical commitments, fact sheets, hiring pipelines, and contact channels, maintained continuously between sessions rather than assembled when a bill drops. Standing infrastructure of that kind is precisely what the seven columns fail to score, and it is the strongest argument yet for an eighth.

Two consequences follow for the ratings above. The Options column understates the position of firms with deep host-county presence, because optionality inside a state — the ability to relocate a project to a friendlier corridor rather than a friendlier state — behaves differently from portfolio optionality across states. And the Conduct column is doing work at two resolutions at once: statewide reputation with the legislature, and corridor-level standing with the host community, which the Wenatchee and Seattle pairing shows can point in opposite directions simultaneously. Both refinements are candidates for the next revision of the master table rather than adjustments made silently here (interpretive, 70–82%).

Why the section sits here rather than in the firm ratings. Externality geography explains a pattern §8 observes: every principal’s hardest terrain includes its home state. Record depth and apportionment produce the same observable through different causes, and only the separation tells an operator which discounts are repairable. The mechanism belongs beside the ratings it explains, not inside them.

X. 💡 Findings

The sorted order does not match the market-cap order — observed at T₀. Google and Microsoft top the field on the axes states actually price (collateral, clean supply, conduct), while the largest spender (Amazon) and the fastest ambition (OpenAI) rank behind them, and the loudest builder (xAI) ranks last. Capital and compute are abundant in this field; authorization capability is scarce, which is TGE-4’s mechanism visible in cross-section. Confidence: 80–88%.

Governance capital is compounding exactly as predicted — consistent-with, not proven. The Seattle withdrawals, the xAI litigation’s citation across unrelated dockets, and the HB 2515 blowback all show conduct converting into price. The ledger test: firms in the ⚠️ and low-Conduct classes should experience measurably longer approvals and more litigation than terms-first firms through 2028 (TLSM-7 restated firm-side).

The dependency graph is the hidden map. Anthropic depends on Fluidstack depends on Google’s credit; OpenAI depends on Oracle depends on Vantage and Crusoe and SoftBank; xAI reportedly folds into SpaceX 🔶. States negotiating with a principal are often actually negotiating with a stack, and the P(honored) that matters is the weakest link’s — a diligence point no state statute yet prices. Confidence: 75–85%. Post-T₀: the Abbott audit directive now requires ownership-structure disclosure across the ERCOT queue — the first state instrument to diligence the stack; the addendum grades it.

Clean portfolios are becoming the passport, and the field is splitting into two leagues. Sixteen Form 1 states — including five of the six highest-certainty jurisdictions — are accessible to roughly half the field. The gas-forward half (Meta, xAI, most neoclouds) is competing for a shrinking accelerate tier that the Authorization Price baseline predicts converts within two sessions of its first campus (Prediction 10). The strategic clock on gas-powered speed is shorter than its practitioners are pricing. Confidence: 75–85%.

The thin benefit line is now a bargaining variable firms choose. Anthropic’s 800-jobs-per-$50B candor and OpenAI’s 25,000-onsite-jobs framing are opposite plays on the same denominator problem; the companion’s loss-aversion mechanics predict the enforceable-guarantee play beats the jobs-headline play as communities learn (TLSM-4, 76–84%), which favors the candid framing over the cycle even though it prices worse today.

🚨 Post-T₀ Addendum — After the Freeze (August 3–14, 2026)

The Atlas recorded the week’s events on the sell side; the addendum below reads the same events from the buy side. Nothing in the body above has been rewritten — T₀ ratings stand, the addendum carries the changes, and every affected claim is graded against what the baseline said before the news.

The Texas audit, read firm-side. Governor Abbott’s August 3 directive ordered PUCT and ERCOT to audit every data center project advancing through the ERCOT interconnection queue — power consumption and onsite generation, water use, tax incentives received, ownership structures, and local-impact mitigation — with non-compliant projects denied connection and the Batch Zero study postponed. ERCOT called the directive an effective pause. The Atlas records the tempo change (BARGAIN → executive audit-gated PAUSE, its ledger’s first row); the firm-side consequences run deeper than the map change.

First, the queue splits by vintage, exactly as §2’s Collateral and Credibility columns predicted the rationing would run. Operators holding executed agreements and completed interconnections hold the appreciated asset; everyone behind them holds an audit obligation. Stargate’s three Texas sites — Abilene operating, Shackelford and Milam in development — now divide along that line, and the division prices OpenAI’s Texas position lower than the T₀ “easiest terrain” cell without touching its operating capacity. Oracle’s strained balance sheet meets a verification screen at precisely the moment its credit story is the field’s most watched. Meta’s position strengthens on a relative basis: self-funded, fully integrated demand is what a verification audit exists to distinguish from the speculative tail, and Meta’s P(honored) is the tier’s highest.

Second — the structurally larger event — a state is now formally diligencing the dependency graph. §9 called the stack behind each principal “a diligence point no state statute yet prices,” and five days after the freeze the second-largest data center state ordered ownership-structure disclosure across a 474 GW queue. The directive verifies rather than scores, and it arrives as a gubernatorial order rather than an RFP or statute, so MC-BP-6 receives adjacent evidence and no settlement credit — the entry’s terms require explicit scoring of conduct or clean-firm supply in an RFP or statute, and grading discipline holds even when the direction flatters the thesis. The neocloud tier carries the sharpest exposure: borrowed bargaining positions are exactly what ownership disclosure surfaces, and the audit converts the tier’s dependency on principal backstops from a private credit fact into a public authorization variable.

Wenatchee, read firm-side. On August 5, the Wenatchee Valley chamber of commerce hosted Microsoft and Sabey Data Centers to a welcoming room, citing twenty-plus years of coexistence in the hydro corridor — the same week Cle Elum, ninety miles south, passed an emergency moratorium three days after its first project surfaced. The Atlas reads the pairing as the local consent market setting Washington’s real price in both directions at once; the firm-side read is what it does to Microsoft’s row. §8 scores Washington as Microsoft’s hardest terrain, and the statewide judgment stands — but the Wenatchee panel shows the terrain is corridor-specific, and incumbency is the one consent asset the conduct discount cannot touch: twenty years of demonstrated benefits lowered the grievance floor in a way no concession package can replicate at entry. Chamber-circuit work by incumbents is bargaining capability in action — the Conduct column operating between sessions, in rooms no register tracks. The specimen also sharpens MC-BP-9’s mechanism: the home-state discount concentrates where the firm-specific record is adversarial (Seattle, HB 2515) and inverts where it is twenty years of quiet performance (Wenatchee), which is the conduct-discount logic operating at sub-state resolution. Interpretive confidence in the corridor-specific reading: 75–85%.

Firm-side response, inside two weeks. Microsoft’s Texas community page — Bexar and Medina counties, Greater San Antonio — carried responsive material within days of the Abbott directive, and the Governor’s office announced on August 13 that Microsoft, alongside Hut 8 and Equinix, had formally committed to comply with the state’s data center standards. The observable is the response time and the channel. A firm that maintains standing per-jurisdiction community infrastructure can answer a gubernatorial directive on its own channel inside a business week, in the host counties, without a press cycle or a lobbyist. Speed of that kind is a capability, and it belongs in the same column as the coalition assets §9 describes — the audit arrived on a Monday and the incumbent’s answer was posted before the second week closed.

What the week did not change. No firm’s composite Position moves. xAI’s row gains nothing and loses nothing — its litigation calendar, not the Texas audit, remains its settlement clock. The sorted-order headline holds: the week’s two events rewarded verified demand, queue vintage, and incumbency conduct, which are the axes the T₀ table already ranked the field on.

XI. 📡 Prediction Register — Frozen at T₀ (July 30, 2026), Entries 9–12 Added August 6 and 8

Event probabilities, settled through named sources, Brier-eligible; interpretive confidence stays in the prose above and never pools with these. Entries MC-FG-1 through MC-FG-8 carry the [T₀] freeze; MC-FG-9 and MC-FG-10 carry [A6] (August 6, 2026); MC-FG-11 and MC-FG-12 carry [A8] (August 8, 2026). The MC-FG namespace is the register native to this paper and is unaffected by the Atlas’s August 6 move to plain-number identifiers; Authorization Price baseline entries are cited here as Prediction n per its Register v2.0 concordance.

Twelve Foresight Simulation Predictions (FSPs) — dated, falsifiable forecasts produced under MindCast’s Cognitive Digital Twin Foresight Simulation discipline — close the paper under the register prefix MC-BP (MindCast Bargaining Power), grouped by theme and numbered in order. Every entry carries the same contract: a percent band, a deadline, a falsifier, and a public settlement source. Interpretive confidence stays in the prose above and never mixes with these. Eight FSPs froze July 30, 2026; the four marked with their dates were added August 6 and 8.

Slot Scarcity and the Price of Authorization

• MC-BP-1 — Authorization value gets a disclosed price. A principal publicly buys queue position or entitled slots from the developer tier at a premium attributed to authorization value. 60–75% · by December 31, 2027. Fails if no disclosed transaction is framed on authorization value. Settles in company filings and transaction press.

• MC-BP-2 — Google’s credit-enhancement model is copied. Another principal backstops a neocloud’s lease or debt obligations. 70–82% · by July 30, 2028. Fails if no comparable backstop is disclosed. Settles in company filings and credit disclosures.

• MC-BP-3 — A Stargate site slips and a host jurisdiction reacts. A Stargate site misses its announced capacity milestone by a year or more, and a host jurisdiction responds with a clawback, renegotiation, or public dispute. 55–70% · by July 30, 2028. Fails if all sites stay within twelve months of milestones, or misses draw no jurisdictional response. Settles in local government records and project announcements.

• MC-BP-4 — Slot-clearing stays scarce (added August 6). In each rolling twelve-month window through the deadline, no more than 60 campuses of 250 MW or larger receive binding authorization across the fifty states, while deduplicated disclosed demand exceeds three times the cleared count. 70–80% · by July 30, 2028. Fails if any window clears more than 60, or demand falls below the 3× multiple. Settles against binding-instrument status per the Atlas's criteria, with interconnection-queue and large-load survey data for the demand side.

Conduct at the Table

• MC-BP-5 — xAI’s turbine record reaches a legal reckoning. xAI loses, settles with material conditions, or is enjoined in the NAACP Clean Air Act action — or removes or permits all contested turbines under pressure. 70–80% · by December 31, 2027. Fails if the suit is dismissed with turbines operating unpermitted and unconditioned. Settles at the federal docket.

• MC-BP-6 — Conduct becomes a scoring criterion. At least one state RFP or statute explicitly scores bidders on governance and conduct record or clean-firm supply. 65–78% · by July 30, 2028. Fails if no state instrument adopts such scoring. Settles in session laws and state RFP documents.

• MC-BP-7 — Governance capability becomes marketing. Two or more principals publicly market authorization or governance capability as competitive differentiation. 80–89% · by July 30, 2028. Fails if fewer than two do so. Settles in company communications and earnings materials. First qualifying specimen logged: Microsoft’s Community-First Infrastructure Initiative, January 13, 2026, with location-specific energy and water disclosure attached; one further principal required to settle. The entry restates a companion Atlas prediction — same mechanism, same band — and the two settle jointly against the same sources, counting once in scoring.

Home-State Geography

• MC-BP-8 — The home-state paradox holds. No principal headquartered in California or Washington breaks ground on a campus of 250 MW or larger in its home state. 75–85% · by July 30, 2028. Fails if any does. Settles in state and local permitting records.

• MC-BP-9 — The home-state conduct discount is real (added August 6). Any principal filing for a 250 MW+ home-state authorization receives materially stricter conditions, longer timelines, or denial relative to its own contemporaneous out-of-state record; if no principal files by the deadline, the entry settles void rather than graded. 65–78% · by July 30, 2028. Fails if a home-state authorization clears at parity with the same firm’s out-of-state record. Settles in permitting records, utility service agreements, and company announcements.

State Bargains and Statute Design

• MC-BP-10 — The clean-for-terms trade goes public. Google or Microsoft announces a state-level partnership explicitly trading new firm-clean generation or transmission funding for siting terms in a strict clean-statute state. 70–82% · by July 30, 2028. Fails if neither announces such a trade. Settles in state agency records and company announcements.

• MC-BP-11 — Washington’s 2027 statute protects ratepayers, not host communities (added August 8). The enacted statute runs strong on ratepayer-protection instruments and comparatively weak on host-community benefits — no mandatory community benefit agreement, no mandatory host-county revenue share, no dedicated local mitigation fund. 68–80% · by June 30, 2027. Fails if the enactment contains any one of the three host-benefit instruments. Settles against Washington’s enacted session law, graded on the Model Code’s host-benefit provisions.

• MC-BP-12 — Host-benefit strength tracks who bears the costs (added August 8). Virginia enacts materially stronger host-benefit provisions than Washington, with Ohio and Georgia falling between — because Virginia’s host and cost jurisdictions overlap and Washington’s do not. 62–75% · by July 30, 2028. Fails if Virginia enacts materially weaker provisions than Washington under comparable load concentration. Settles against enacted law in the four states, graded on the same Model Code provisions.

Register notes, for auditors rather than readers: entries are Brier-eligible event predictions settled through the named sources; the MC-BP namespace is native to this paper — entries froze under a prior working prefix and sequence, and were renamed and renumbered by theme on the paper’s retitle with content, bands, deadlines, and falsifiers unchanged — and is unaffected by the Atlas’s August 6 move to plain-number identifiers; Authorization Price baseline entries are cited as Prediction n per its Register v2.0 concordance; the paired Atlas entries behind MC-BP-7 are its Predictions 27 and 28, with no credit taken on 28 because Microsoft’s framing answers community concern rather than naming misinformation.

XII. ✅ Verification and Maintenance

Facts verified by web search July 30, 2026: capex figures from Q1-2026 earnings compilations (FT via Tom’s Hardware; Data Center Frontier-class trackers); Stargate site status from Epoch AI’s April 2026 satellite survey, OpenAI’s own announcements, and CNBC’s June 1 Michigan groundbreaking coverage; the xAI record from the SELC/Earthjustice litigation file, the February 2026 notice of intent, and July 2026 investigative reporting on the 59-turbine count; Anthropic’s program from its November 12, 2025 announcement and DCD’s site-level analysis; Google’s backstop structure from DCD’s financing reporting.

Firm-side primary sources: Microsoft Local jurisdiction pages (local.microsoft.com), specifically the Washington page (Quincy, Malaga, East Wenatchee; modified June 24, 2026) and the Texas page (Bexar and Medina counties; modified August 11, 2026, carrying TX_AbbottAug2026.pdf); Community-First Infrastructure Initiative announcement (blogs.microsoft.com, January 13, 2026); Understanding energy use at Microsoft datacenters (local.microsoft.com, published March 30, 2026, modified August 14, 2026), carrying the cost-causation definition, PUE disclosure, and clean-portfolio figures cited above; FY25 Datacenter Environmental Data Fact Sheet; per-jurisdiction datacenter operations fact sheets (twelve US states); Datacenter Community Pledge (June 2, 2024).

Post-T₀ sources (August 3–6, 2026): Abbott letter to PUCT/ERCOT via Argus Media, KERA News, ABC13 Houston, and TechCrunch, with the ERCOT effective-pause statement via Fox News; Wenatchee Valley Chamber of Commerce Coffee & Commerce panel post (LinkedIn, August 5, 2026); Seattle Times on Cle Elum (August 5, 2026); Financial Times 2026 hyperscaler capex reporting for the $725B→$745B movement noted in the opening.

Single-source items carry 🔶: the SpaceX-xAI ownership claim, the Stack/BorderPlex $165B figure, and Related Digital’s posture. Highest-churn rows: OpenAI (site plans moved once already), Oracle (credit narrative), xAI (litigation calendar), the Texas cells (audit-standard publication is the trigger), and every capex figure at the next earnings cycle. Refresh quarterly at minimum; refresh the xAI, Stargate, and Texas rows on event.

The scoring axes inherit their definitions from The AI Data Center Authorization Price §I–III, the identifier definitions from the Atlas’s Appendix H, and the provision vocabulary from the Model Data Center Authorization Code; this paper is the firm-side half of the same evidence layer; its twelve Foresight Simulation Predictions are its simulation output, graded through the register above.

XIII. Conclusion

Bargaining power in the authorization market is a manufactured asset, not an inherited one. Seven capabilities decide who wins scarce slots, conduct multiplies or discounts the whole bundle, and the master table’s medals fall out of that arithmetic: Microsoft and Google lead not because they are largest but because they carry collateral, clean supply, and governance simultaneously, while the announce-first, gas-forward half of the field loses across every tempo for what is ultimately one weakness wearing four costumes.

The field is splitting into two leagues on a clock. Sixteen clean-statute states already price the power portfolio before negotiation begins, the accelerate tier is converting, and clean-portfolio depth is compounding from a regional advantage into the field-wide passport. Firms holding the passport bank terms; firms without it are racing a shrinking map — and the strategic clock on gas-powered speed runs shorter than its practitioners are pricing.

States reading these ratings should notice the reciprocal: the same table that tells a firm where it can bargain tells a state which bidders can actually deliver. Conduct records, collateral capacity, and demand credibility are diligence variables a state RFP can score today, and the first statutes to score them will set the template the rest copy.

Twelve Foresight Simulation Predictions turn the analysis into a wager — on slot scarcity, on conduct converting into a scored criterion, on the home-state paradox, on whose statutes protect whom. Each carries its band, deadline, falsifier, and settlement source, and the ratings above should be judged by how that register grades. The companion papers price the states’ side of the bargain; the record from here forward prices ours.

See MindCast Data Center Regulatory Economics and Innovation Live-Fire Mission

Executive Summary

Four sponsor-backed colocation and wholesale platforms moved toward public price discovery inside six weeks, and cohort membership stays fixed at those four for every prediction that follows. Vantage Data Centers, Switch, CyrusOne, and DayOne each took a step toward a listing or a sale between early July and mid-August 2026.

Financial coverage has read the wave as AI demand lifting valuations. A closer reading of the calendar points somewhere else.

Vantage reduced the immediate regulatory uncertainty surrounding its largest Texas campus on August 12 and tested the transaction market on August 13. Wisconsin regulators had already rewritten the tariff governing Vantage’s second-largest campus on April 24, attaching a fifteen-year commitment and a collateral requirement keyed to the credit rating of the customer taking service under the tariff.

Sponsors are testing whether public markets will capitalize development pipelines at the moment their largest host jurisdictions began attaching explicit prices, terms, and delays to authorization. The registration statement required to complete a listing is the document most likely to force material authorization terms into public view.

Three consequences follow. Megawatts stop behaving as fungible assets. A public listing operates as a fourth authorization proceeding alongside government, utility, and customer. And the first registration statement places authorization exposure in a liability-backed public document that legislators and investors read at the same time.

What MindCast Forecasts

Section VIII registers forty-three dated calls, each with a falsifier and a named settlement source. Ten test the central mechanism, and the eight headline calls below summarize the argument.

• Capacity disclosure arrives first — 80–90%. The first filing quantifies three or more capacity states and names customer concentration.

• A state or regulatory regime appears by name — 72–82%. Authorization exposure enters securities disclosure jurisdiction by jurisdiction.

• A completed Vantage IPO is the minority outcome — 33–43%. Minority investment or recapitalization runs 24–34%, no major transaction completed 19–29%, control sale 5–13%.

• The first cohort listing prices below its anchor — 60–70%. Public markets discount pipelines that private rounds capitalized whole.

• Texas codifies its standards by end-2027 — 75–85%. Three more states adopt financial-assurance or cost-protection rules by end-2028 — 70–80%.

• A state record cites a cohort filing within twelve months — 55–65%. Securities disclosure reaches legislative drafting.

• Equity research names legislative or siting action as a valuation factor — 65–75%. Political risk starts moving price targets.

• A listed developer’s shares move on an adverse authorization event — 50–60%. Three points of underperformance against a named benchmark over five trading days.

Every forecast carries a deadline, a falsifier, and a public settlement source, so a reader can grade the analysis against the record rather than accept it on trust.

The sections that follow supply the mechanism behind the numbers: what two Texas actions and one leak did inside nine days, how nineteen states treat the same portfolio under four different authorization regimes, why capital markets now function as a fourth authorizer, and what a single registration statement will expose that three legislative sessions have not.

I. Two Texas Actions and an IPO Leak, Nine Days Apart

Three public actions occurred inside nine days, each reported separately and none connected to the others in the coverage that followed. Placing all three on one timeline changes what the initial public offering means.

August 3. Governor Greg Abbott directed the Public Utility Commission of Texas and ERCOT to audit every data center project in the interconnection queue before any project advances. ERCOT held roughly 474 gigawatts of connection requests, about ninety percent of them data centers, and more than five times the state’s record peak demand. Pending data center interconnection requests hold until each project completes the audit.

August 12. Abbott announced that Core Scientific, SB Energy, and Vantage Data Centers had committed to comply with the state standards. Vantage holds Frontier in Shackelford County — 1,200 acres, 1.4 gigawatts, ten buildings, and more than $25 billion of announced investment. Frontier is the largest campus in the company’s global portfolio.

August 13. Reuters reported that Vantage is exploring an initial public offering that could raise roughly $10 billion at a valuation near $100 billion, with a sale or minority stake sale also under consideration. Deliberations remain preliminary and no formal process has launched.

Vantage moved its largest Texas campus from open regulatory exposure to declared compliance on August 12, with the audit still ahead, and Reuters reported transaction deliberations the following day.

Sellers reduce regulatory uncertainty before testing a market. The incentive operates on any seller holding a gigawatt behind a paused queue.

Wisconsin supplies the second half of the calendar. The Public Service Commission of Wisconsin approved the We Energies Very Large Customer tariff on April 24 after rewriting it — extending the minimum initial term to fifteen years, lowering the eligibility threshold from 500 megawatts to 100 megawatts, removing the capacity-only option, and ordering revisions against transmission cost-shifting. Vantage’s Lighthouse campus in Port Washington, a $15 billion development approaching a gigawatt and scheduled for 2028, takes service under that tariff.

Credit terms sit at the center of the Wisconsin order. Customers that cannot meet the A- credit standard must provide cash, letters of credit, or other required financial assurance, and We Energies asked the Commission in June to revisit the requirement, arguing the standard burdens data center operators.

Vantage is not moving alone. Switch filed confidentially on August 7 for a listing that could arrive in the fourth quarter, while CyrusOne targets 2027 and DayOne could go public next quarter.

Four platforms seeking public price discovery inside one window is not four independent judgments about capital need. Four platforms doing so in the same season that major host states are rewriting authorization terms marks the convergence of two cycles: unprecedented capital demand, and the explicit repricing of power access.

A fifth issuer moves on a faster clock and sits outside the cohort. London-based Nscale told prospective investors it holds roughly $51 billion in contracted revenue and could list in the United States as early as September, with sites developing in West Virginia, Norway, and a 40-megawatt colocation position in Madison, North Carolina. Vertical integration across power, data centers, GPUs, and software places Nscale nearer CoreWeave than nearer Vantage, and a GPU-cloud issuer discloses customer concentration and compute contracts where a colocation platform discloses capacity states and interconnection position. Exclusion from the cohort follows from the disclosure difference rather than from timing, and the predictions below do not settle against an Nscale filing.

Convergence sets the stakes. Authorization terms now determine what the listing wave must disclose and what investors must price.

II. Who Should Read This and Why

Seven groups face different decisions on the same set of facts. Each one takes something different from the analysis that follows, and each is named here with the decision at stake rather than the label alone.

Institutional investors and allocators. Any bid on a data center listing is a bid on capacity that has not yet cleared authorization. Announced gigawatts appear in marketing materials long before the permits, tariffs, and interconnection agreements that convert them into revenue.

Section III supplies a six-class scale for separating capacity that produces revenue from capacity that produces press releases, and states plainly what the scale does not measure. Section IV maps the cohort’s roughly nineteen states onto four recurring authorization patterns and names the two MindCast instruments that make them comparable. Section VII names the five disclosures that would resolve the question, and Groups Six and Seven of Section VIII trace how state legislative sessions, moratoria, and local opposition reach offering execution, analyst price targets, and share prices.

Government affairs and site selection teams at hyperscalers and developers. Capital structure has become an authorization variable, and authorization conduct has become a valuation variable. Both directions are new, and both change how a government affairs function should be resourced.

Section VI explains how credit-keyed collateral rules connect capital structure to authorization, without assuming that public status satisfies the requirement. Section V explains why a firm’s bargaining record across many statehouses now carries a valuation consequence rather than a reputational one, because public markets price the platform rather than the site. A concession made in one state may become evidence of platform-wide bargaining behavior for investors evaluating a campus in another.

Developers should assemble a portfolio-wide authorization record before entering any transaction process, and rank jurisdictions by computability rather than by nominal leniency. Site-selection teams should treat tariff certainty, transferability, collateral exposure, and time to firm power as capital-market variables rather than post-selection regulatory detail.

Hyperscale customers occupy a different seat at the same table. Material concentration, load commitments, credit support, and project dependencies enter public disclosure once they affect platform value, which may strengthen a customer’s apparent credit while narrowing the confidentiality around its bargaining position.

State lawmakers and legislative staff. Legislatures have spent two sessions attempting to compel data center disclosure and have been met with confidentiality agreements and the absence of any authority requiring a private developer to state what it fears.

Section VII identifies the portions of that inventory materiality may force into public disclosure, and the portions securities law will leave untouched. Section VI explains why credit-keyed ratepayer protections and corporate disclosure push in the same direction, which gives drafters a rare alignment to work with. Lawmakers in states holding completed data center capacity should read Sections III and IV first. The six-class scale explains why a state can host a finished campus and still lose the next one, and the regime table shows which neighboring states now produce a computable obligation.

Utilities, commissions, and public power districts. Cost allocation for large loads is being settled state by state, and the settlements are diverging sharply. Wisconsin regulators wrote a fifteen-year term with collateral posted against credit rating. Texas paused an entire interconnection queue pending audit. Several states have written nothing at all.

Section I sets out what the two most aggressive states did and when, and Section IV places both alongside the contested and unwritten regimes the same cohort operates under. Section VI explains why terms that look restrictive on their face can lower a developer’s cost of capital by making obligations legible, and why the absence of any defined term does the opposite.

Commission staff should reconcile disclosed campus schedules, customer concentration, and capacity classifications against utility load forecasts, interconnection assumptions, and financial-assurance calculations. A material gap between a securities filing and a utility docket identifies where ratepayer exposure has been understated or project maturity overstated.

Consumer-owned utilities and public utility districts should note that most large-load tariff frameworks now under debate reach investor-owned utilities first, leaving the fastest-growing rural load outside the framework entirely.

Sponsors, issuers on the listing track, and their transaction advisers. A registration statement converts every authorization negotiation into a disclosure decision, and the disclosure decision arrives before pricing rather than after it. Sponsor liquidity and capital recycling may decide which route the transaction takes.

Section III supplies the capacity classification investors will apply whether or not a filing volunteers it, and Group Two of Section VIII sets a benchmark for what peers will disclose. Section VII names the five disclosures underwriters and counsel will argue over, including the one most often deferred — how much of an offering funds construction, debt reduction, collateral support, or sponsor liquidity.

Advisers should test early whether a listing, sale, or recapitalization triggers change-of-control consent, assignment restrictions, collateral resets, tariff eligibility changes, or incentive clawbacks across the portfolio. Authorization does not always travel with ownership.

County councils and local data center strategists. A developer negotiating in front of public shareholders behaves differently from one negotiating in front of two private sponsors, and the difference shows up at the counter where local agreements get made.

Section V sets out the predicted direction of that change, and Entry 22 in Section VIII tests whether a listed developer takes on recurring, jurisdiction-level disclosure of power, permitting, and community exposure. Local officials should use the first filing to compare a developer’s concessions, timelines, and risk disclosures across jurisdictions before approving incentives, infrastructure commitments, or community agreements.

Ratepayer advocates and community organizations gain a new evidence source rather than a complete accountability record. Securities filings disclose what threatens investors, while legislation, dockets, and local agreements remain the instruments that disclose what threatens ratepayers and communities.

Credit investors, lenders, and rating agencies. A public listing broadens the equity base without eliminating project-level tariff obligations, collateral requirements, or construction exposure.

Section VI identifies the question that governs credit rather than equity: not whether the parent becomes public, but which legal entity takes service, posts financial assurance, and carries the long-dated power commitment. Entries 47, 52, 56, and 58 test whether authorization quality enters formal valuation and credit methodology.

Every reader arrives at the same underlying claim from a different door: authorization has stopped functioning as a permitting cost and started functioning as a valuation input. The sections that follow show the mechanism, then register the falsifiable consequences.

III. Six Kinds of Capacity, Six Different Values

Private infrastructure valuation aggregates. A single headline gigawatt figure can absorb operating capacity, capacity under construction, capacity holding a power reservation, capacity sitting in an interconnection queue, and capacity attached only to controlled land.

Sponsors and their lenders already know which is which, because diligence in a private transaction reaches the underlying agreements. Public investors cannot reliably reconcile the portfolio without issuer disclosure.

Six classes carry genuinely different value. The distinction between them is legal and physical rather than accounting, which is why no financial statement currently discloses it.

Vantage’s own portfolio spreads across the full range. The Quincy campus in Grant County, Washington sits in class one, at 89 megawatts across three buildings and roughly $1 billion, completed in 2023 on Grant County Public Utility District power. Frontier delivers its first building in the second half of 2026, placing one increment near class two and the remaining nine buildings in classes three and four. Lighthouse completes in 2028 under a tariff whose collateral terms remain contested at the Commission.

The six-class scale measures authorization and deliverability rather than commercial quality. Investors should apply it alongside customer contract term, counterparty credit, cancellation rights, and revenue concentration. An authorized megawatt without durable customer revenue and a contracted megawatt without deliverable power fail for different reasons, and a single capacity total conceals both failures.

Announced gigawatts and revenue-producing gigawatts are separated by exactly the legal and physical steps MindCast measures across all fifty states. The right question for an investor is not how many megawatts a developer has announced. The right question is how many have cleared authorization, and what clearing the remainder will cost.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

IV. Nineteen States, Four Authorization Regimes

Four platforms do not share a regulatory problem. Across roughly nineteen states, the cohort operates, develops, or pursues campuses under materially different authorization conditions.

Vantage operates or develops across Texas, Wisconsin, Washington, Ohio, Georgia, California, Virginia, Arizona, Indiana, and Nevada. Switch runs campuses in Nevada, Michigan, Georgia, and Texas, and announced a 382-acre Pennsylvania campus in April 2026. CyrusOne holds facilities in Virginia, Texas, Arizona, Ohio, Illinois, Iowa, North Carolina, New Jersey, New York, Connecticut, Kentucky, and Washington. DayOne brings a largely Asia-Pacific and European portfolio toward a proposed United States listing.

Four recurring patterns illustrate how one portfolio meets materially different state conditions, and each prices differently.

Local institutions carry the load wherever no state instrument exists. Siting approvals for Switch’s Nevada and Pennsylvania campuses run through county and municipal bodies, while power service and interconnection remain subject to separate utility and grid processes. Switch has already committed publicly to funding its own power infrastructure without passing costs to consumers, a concession volunteered ahead of any filing requirement.

Comparing four patterns across nineteen states requires a common scale, and MindCast built two. The 50-State Authorization Price Atlas grades every state on the conditions a developer actually faces. The Model Data Center Authorization Code supplies the strictness scale that converts scattered risk-factor language into one comparable number.

Applied to a registration statement, the two instruments answer a question no filing answers on its own: whether a portfolio concentrates in states that can be priced or in states that cannot.

Washington supplies the sharpest case, and not because of a single company. Two of the four platforms hold Grant County capacity — Vantage at its Quincy campus and CyrusOne at PNW1 — served by a consumer-owned utility outside commission jurisdiction, in a state with no statewide large-load instrument. Neither campus sits inside a state commission tariff docket, and both depend on public-power instruments outside commission jurisdiction.

Investors reading a multi-state filing will therefore encounter jurisdictions that produce a computable obligation and jurisdictions that produce silence. Silence is the harder one to price, and the Atlas exists to convert it into a grade.

V. Governments, Utilities, Customers, and Now Capital Markets

Four institutions authorize an AI data center. Each one can stop a project cold, and a project that satisfies three of them still does not get built.

Governments authorize construction and operation through statute, ordinance, and permit. Utilities and their commissions authorize access to power through tariff, interconnection agreement, and cost allocation. Customers authorize revenue through contract. Capital markets authorize the next development cycle by pricing what the first three have done.

One decisive difference separates the fourth authorizer from the other three. Governments, utilities, and customers each evaluate a single project in a single jurisdiction, on its own schedule, behind its own confidentiality terms. Capital markets evaluate the whole portfolio at once, across every jurisdiction, on a single day, and publish the answer as a number the world can read.

An initial public offering therefore operates as the fourth authorization proceeding, conducted in a different forum under different rules. Underwriters run discovery. Counsel drafts disclosure under liability. Investors render a verdict on the developer’s complete authorization record rather than on any one campus.

Developers that bargained well across many statehouses should receive a different multiple from developers that bargained expediently in a few. A concession made in one state may become evidence of platform-wide bargaining behavior for investors evaluating a campus in another, because public markets price the platform rather than the site.

VI. Why Power Approval Now Changes Valuation, Not Only Cost

MindCast’s baseline work established that every project pays an authorization price, paid in cost, obligation, and delay rather than in permission alone. The credit analysis that followed established that authorization uncertainty travels through the discount rate, the borrowing cost, the schedule reserve, and the covenant package.

A public listing advances the mechanism one step further. Authorization uncertainty stops behaving like a cost and starts behaving like a multiple.

Compounding explains the difference. A cost is subtracted once from one project. A multiple applies to every future megawatt the platform expects to build, so a developer with a demonstrably stronger authorization record earns a premium on capacity it has not yet acquired.

Wisconsin shows the mechanism running in the opposite direction. A fifteen-year minimum term converts a tariff into a long-dated liability with a known counterparty. A collateral requirement keyed to credit rating converts the developer’s capital structure into an authorization input.

A public listing would broaden Vantage’s equity base, add audited financial reporting, and give investors a continuously observable measure of platform credit quality. Whether those changes lower project-level collateral costs depends on which entity a tariff names as the responsible customer, which entity posts financial assurance, and what that entity’s rating actually is. Public status alone resolves none of the three.

Credit-keyed tariff requirements connect corporate finance to authorization. A commission writing collateral rules is writing capital-structure policy, whether or not it intends to.

Legislators should note where the incentive points. Credit-keyed ratepayer protections give a developer reason to improve financial transparency and strengthen the entity standing behind the load. Ratepayer protection and corporate disclosure push in the same direction, which is an unusual and useful alignment for anyone drafting in 2027.

A second lesson runs the other way and matters more for states that have written nothing. A developer can compute a Wisconsin electricity bill fifteen years forward, because the Commission set a term, a threshold, and a collateral rule. A developer cannot compute a bill in a state with no large-load framework at all, and an uncomputable cost gets priced as risk rather than as a number.

Washington illustrates the gap. Vantage completed its Quincy campus, publicly credited the rural tax incentive and its utility partnership, and placed its next 2.4 gigawatts in Texas and Wisconsin. Washington’s large-load tariff legislation passed the House in 2026 and died in the Senate, leaving the state with a tax preference and no statewide instrument for making future large-load obligations legible.

Texas and Wisconsin each handed a developer terms it could compute. Washington hands a developer a tax preference and a blank space where the obligations belong.

Uncertainty costs a state more capacity than strictness does. MindCast carries the proposition as a working hypothesis and tests it through the Group Five entries in Section VIII, where adoption and enactment records settle it against the 2027 and 2028 legislative record.

VII. What an SEC Filing Will Reveal That State Proceedings Have Not

MindCast’s earlier work on missing dockets made a structural argument — material authorization exposure frequently sits outside the proceeding observers are watching. The current IPO wave relocates the docket to a venue no state government monitors.

A registration statement requires disclosure of material risks under securities-law liability. For a development platform whose value rests on a small number of enormous campuses, materiality can force authorization exposure into jurisdiction-specific language. Moratoria, tariff reallocations, audit-gated interconnection pauses, ordinance denials, and cost-allocation rulings stop being operational friction described in a press release. Any such instrument that materially affects portfolio value, project timing, or expected revenue may enter the filing, require updating as the exposure changes, and expose the issuer if stated wrongly.

Five disclosures would carry the most information for every audience named in Section II:

• Portfolio split across the six capacity classes. How much of the announced gigawatt total is energized, and how much is intention.

• Firm power versus queue position. How much announced capacity holds deliverable, contracted power rather than an interconnection number.

• Counterparty concentration. Which customers carry the contracted revenue, and what happens to the platform if one of them slows.

• Named jurisdictions. Which states the issuer identifies as sources of authorization risk, and which instruments it names.

• Capital destination and obligated entity. How much of the offering funds construction, debt reduction, collateral support, or sponsor liquidity; which legal entities hold the campuses and the power commitments; and whether ownership changes require consent, reassignment, or renewed financial assurance.

Capital destination deserves as much attention as capacity. A ten billion dollar offering that funds construction tells a state one thing about project maturity; the same offering funding sponsor liquidity tells a state something else entirely, and the two are indistinguishable from the headline number.

Foreign issuers carry the mechanism across borders and compress it. Singapore-based DayOne raises American capital to build in Malaysia, Indonesia, Finland, and Spain, while Nscale would register American disclosure covering West Virginia and Texas authorization terms alongside Norwegian ones. A single registration statement written for United States investors then states, in one document and one language, what a developer computes in West Virginia against what the same developer computes in Norway. State authorization terms enter the comparison as line items rather than as background, and the comparison reaches investors who never read a commission docket.

Authorization transferability is the question hiding underneath. Wisconsin keyed collateral to the credit rating of the customer taking service, and Texas requires developers to disclose project ownership and controlling interests. A listing, sale, or recapitalization that moves the obligated entity can therefore reopen collateral, eligibility, and incentive terms that everyone treated as settled.

States should ask a plain question of any developer entering a transaction: which entity carries the obligation after the deal closes, and does the obligation survive the change.

Named jurisdictions deserve particular attention in statehouses. Legislatures have largely failed to compel data center disclosure, blocked by confidentiality agreements and by the absence of any authority requiring a private developer to state what it fears. A filing may produce parts of that inventory voluntarily and under liability, though disclosure will follow securities-law materiality rather than the information needs of legislators or communities.

Securities disclosure reaches a defined perimeter. Filings arrive after siting and tariff decisions are made, cover listed companies only, and describe risk to investors rather than obligations to communities. Legislation governs everything outside that perimeter.

Inside the perimeter, securities law is positioned to deliver more data center transparency in 2027 than data center law has delivered in three sessions. Legislators and county officials should read the first filing the week it appears and treat the contents as drafting evidence.

VIII. Forty-Three Dated Predictions

Forty-three forward claims, dated August 13, 2026. Each carries an explicit falsifier, a named settlement source, a deadline, and a confidence band with its class label. Predictions are numbered 1 through 43 for this publication and correspond to master MindCast register numbers 17 through 59.

Predictions run in seven groups tracing one loop: how the transaction resolves, what the filings disclose, where valuation lands, how later listings behave, how states respond, how legislative sessions and local opposition enter the record, and how all of it returns as price.

Three standing rules govern the register and apply to every prediction below.

Activation. A conditional prediction whose trigger never occurs remains unactivated and is excluded from scoring rather than counted correct or incorrect. Filing-dependent predictions settle only against a registration statement that actually becomes public.

Class. Event-class claims settle against an external record. Interpretive claims carry a stated matching rule and stay out of any probabilistic aggregate.

Priority. Ten predictions marked Primary test the mechanism this note argues. Predictions marked Secondary test jurisdiction and actor behavior around that mechanism and are scored separately.

Independent model runs on the same input set agreed on the governing mechanism and the regime classification. Where independent runs agreed on the mechanism but assigned different route probabilities, the register preserves the full cross-model confidence envelope and reports the equal-weight central estimate separately. Cross-model agreement does not independently raise confidence.

Group One — Transaction Path

Seven predictions covering whether a transaction happens, in what form, and in what order across the cohort. Coherence constraint applied: the probability of launching a qualifying process is set at or above the probability of completing any transaction.

1 — Route Decomposition [Primary] Vantage’s transaction path resolves into one of four exclusive routes by the deadline: a completed and priced initial public offering at 33–43% central, 30–45% envelope; a minority investment or private recapitalization at 24–34% central, 22–36% envelope; no major transaction completed at 19–29% central, 18–31% envelope; a control sale at 5–13% central, 4–15% envelope. Event class. Routes are mutually exclusive and jointly exhaustive, and central-band midpoints sum to 100. Envelope bands span the full cross-model range and do not sum to 100. Scoring: categorical Brier or logarithmic score across the four routes, since a low-probability route occurring does not by itself defeat a distribution. Structural falsifier: the observed outcome cannot be assigned exclusively to one of the four routes. Settlement: SEC filings and company announcements. Deadline: December 31, 2027.

2 — Formal Process [Secondary] Vantage announces or launches a formal transaction process — a public or confidential registration filing, a mandate announcement, or a confirmed sale process. Confidence: 75–85%, event class. Falsifier: no announced or launched process by the deadline. Settlement: company announcements and SEC filings. Deadline: December 31, 2027.

3 — Listing Completion [Secondary] Vantage prices an initial public offering. Confidence: 33–43%, event class. Prediction 3 restates the initial-public-offering component of Prediction 1 and receives no separate score. Falsifier: no Vantage initial public offering prices on or before the deadline. Settlement: SEC filings and pricing announcements. Deadline: December 31, 2027.

4 — Sequence [Secondary] Switch prices, or a Switch registration statement becomes publicly available, before any Vantage registration statement becomes publicly available. Confidence: 70–80%, event class. Falsifier: a Vantage registration statement becomes public first, or neither becomes public by the deadline. Settlement: EDGAR. Deadline: December 31, 2027.

5 — Path Divergence [Secondary] At least one of the four cohort members resolves through a private sale or minority transaction rather than pricing a listing. Confidence: 55–65%, event class. Falsifier: all four either price a listing or take no transaction action by the deadline. Settlement: company announcements and SEC filings. Deadline: December 31, 2027.

6 — Conversion Cap [Secondary] No more than two of Vantage, Switch, CyrusOne, and DayOne complete United States listings. Confidence: 65–75%, event class. Falsifier: three or more of the four complete United States listings by the deadline. Settlement: SEC filings and pricing announcements. Deadline: December 31, 2027.

7 — Offering Size [Secondary] Conditional on Vantage pricing an offering, gross proceeds fall below $10 billion.Confidence: 65–75%, event class. Falsifier: gross proceeds at or above $10 billion. Unactivated if no Vantage offering prices. Settlement: final prospectus. Deadline: December 31, 2027.

Group One tests whether the wave arrives. The remaining groups test what the wave produces once their filing, listing, or pricing triggers activate.

Group Two — Disclosure Content

Nine predictions testing what a registration statement reveals about authorization. Predictions 8 through 10 settle against the same document and count as one line of evidence.

8 — State or Regime Identification [Primary] The first registration statement to become publicly available among the four cohort members identifies at least one state or named regulatory regime as materially affecting power delivery, construction timing, cost allocation, or project economics. Confidence: 72–82%, event class. Falsifier: the filing identifies no state or named regime in connection with those categories. Settlement: EDGAR. Deadline: December 31, 2027.

9 — Named Instrument [Secondary] The same filing names a specific authorization instrument — an identified tariff docket, commission order, state audit directive, moratorium, or interconnection pause. Confidence: 40–50%, event class. Falsifier: the filing names no specific instrument or proceeding. Settlement: EDGAR. Deadline: December 31, 2027.

10 — Two-State Appearance [Secondary] Texas and Wisconsin both appear in the first publicly available Vantage registration statement in connection with power delivery, tariff, interconnection, regulatory, or development risk, counted anywhere in the document. Confidence: 45–55%, event class. Falsifier: either state is absent from those contexts. Unactivated if no Vantage filing becomes public. Settlement: EDGAR. Deadline: December 31, 2027.

11 — Capacity Separation [Primary] The first cohort registration statement quantifies at least three capacity states — drawn from operating, under construction, contracted, committed, and planned — and identifies material customer concentration. Confidence: 80–90%, event class. Falsifier: fewer than three capacity states quantified, or customer concentration omitted. Settlement: EDGAR. Deadline: December 31, 2027.

12 — Capacity-State Standardization [Secondary] At least two of the first three public cohort filings separately quantify operating, under-construction, and contracted or committed capacity. Confidence: 75–85%, event class. Falsifier: fewer than two of the first three filings quantify all three states separately. Unactivated if fewer than three cohort filings become public. Settlement: EDGAR. Deadline: December 31, 2027.

13 — Firm-Power Limit [Secondary] The first cohort filing distinguishes broad development stages without identifying firm, deliverable power separately for every material campus. Confidence: 75–85%, event class. Falsifier: the filing identifies firm, deliverable power on a campus-by-campus basis for every campus it describes as material, or the filing fails to distinguish broad development stages at all. Settlement: EDGAR. Deadline: December 31, 2027.

14 — Portfolio-Table Limit [Secondary] No cohort registration statement discloses a jurisdiction-by-jurisdiction megawatt table covering the full portfolio. Confidence: 65–75%, event class. Falsifier: any cohort filing discloses such a table. Unactivated if no cohort filing becomes public, and not settled correct by absence of filings. Settlement: EDGAR. Deadline: December 31, 2027.

15 — Structural Concentration [Secondary] At least one cohort registration statement discloses a related-party, common-control, or concentration risk arising from overlapping sponsor, customer, or joint-venture relationships across the AI compute stack. Confidence: 45–55%, event class. Falsifier: no cohort filing discloses such a relationship. Unactivated if no cohort filing becomes public. Settlement: EDGAR. Deadline: December 31, 2027.

16 — Regulator-Forced Expansion [Primary] SEC staff issues at least one comment letter to a cohort issuer requesting expanded disclosure on power availability, interconnection status, energization timing, or customer concentration.Confidence: 55–65%, event class. Falsifier: no such comment letter appears in released correspondence for any cohort issuer. Unactivated if no cohort registration statement becomes public. Settlement: UPLOAD and CORRESP filings on EDGAR. Deadline: 18 months after the first cohort registration statement becomes public.

Predictions 13 and 14 mark the perimeter deliberately. Securities disclosure will surface capacity staging and named jurisdictions while leaving campus-level firm power unresolved, and Group Two establishes both halves.

Group Three — Valuation

Three predictions testing whether public markets price authorization quality rather than merely disclose it.

17 — Cohort Haircut [Secondary] The first cohort listing to price does so at an implied valuation below the highest pre-filing figure reported for that company. Confidence: 60–70%, event class. Falsifier: pricing at or above the highest reported pre-filing figure, measured on the same basis the report used. Unactivated if no cohort listing prices. Settlement: pricing announcement compared against the reported anchor. Deadline: December 31, 2027.

18 — Vantage Haircut [Secondary] Conditional on Vantage pricing an offering, the headline valuation prices below $100 billion. Confidence: 65–75%, event class. Falsifier: pricing at or above $100 billion on the basis stated below. Settlement basis stated here: comparison runs on equity value at the offer price against the $100 billion figure read as equity value; where the filing establishes the reported figure was enterprise value, settlement runs on enterprise value at pricing. Unactivated if no Vantage offering prices. Settlement: final prospectus and pricing announcement. Deadline: December 31, 2027.

19 — Authorization-Linked Valuation Split [Primary] Public-market analysis distinguishes operating-yield platforms from development-option platforms and ties the differential explicitly to capacity status, deliverable power, authorization exposure, or jurisdictional risk. Confidence: 70–80%, interpretive class. Matching rule stated here: at least two published sell-side initiation or update notes covering different cohort issuers apply materially different valuation frameworks or capacity discount assumptions, and each states capacity status, power deliverability, authorization exposure, or jurisdictional risk as a stated basis for the difference. Falsifier: no two such notes appear, or the differential rests solely on growth, leverage, scale, or customer mix. Unactivated if fewer than two cohort issuers are publicly listed. Settlement: published equity research. Deadline: December 31, 2028.

Prediction 19 carries the valuation half of the thesis. A differential tied explicitly to authorization quality is authorization uncertainty behaving as a multiple rather than a cost.

Group Four — Later Listings and Market Structure

Four predictions testing whether the first filing sets a template and how the cohort behaves afterward.

20 — Post-Listing Disclosure [Primary] The first cohort member to complete a listing updates, repeats, or incorporates by reference jurisdiction-specific discussion of power availability, permitting, interconnection, cost allocation, or community opposition in at least one of its first two scheduled post-listing periodic reporting cycles. Eligible documents include Forms 10-Q, 20-F, and 6-K earnings materials, or the equivalent required periodic report. Confidence: 65–75%, event class. Falsifier: neither of the first two reporting cycles carries such discussion in any form. Unactivated if no cohort listing completes. Settlement: EDGAR. Deadline: completion of the second scheduled reporting cycle after the first cohort listing prices.

21 — Template Formation [Secondary] The second and third cohort registration statements share at least three of four authorization risk categories with the first: power availability or interconnection, cost allocation or tariff exposure, permitting or siting, and community or political opposition. Confidence: 60–70%, interpretive class. Matching rule stated here: a category counts as shared where the filing devotes at least one dedicated risk-factor heading or equivalent discrete discussion to it. Falsifier: fewer than three categories shared across the three filings. Unactivated if fewer than three cohort filings become public. Settlement: EDGAR. Deadline: December 31, 2027.

22 — Contagion [Secondary] Within six months after the first cohort listing prices, at least one remaining cohort member changes its expected timing, valuation, offering size, or transaction route against the baseline in effect on the pricing date. Confidence: 65–75%, event class. Baseline set at pricing: the most recent publicly reported timing, valuation, size, and route for each remaining member as of that date. Falsifier: no remaining member’s reported plan changes against that baseline within six months. Unactivated if no cohort listing prices. Settlement: company announcements, SEC filings, and reported changes attributed to named sources.

23 — Weak Pricing Substitution [Secondary] Conditional on the first cohort listing pricing below its marketed range or falling at least ten percent across its first thirty trading days, at least one remaining cohort member delays a listing or pursues private capital within six months. Confidence: 70–80%, event class. Falsifier: no remaining member delays or pursues private capital within six months of the trigger. Unactivated if the trigger condition never occurs. Settlement: pricing records, trading data, and company announcements.

Group Four determines whether one filing produces a standing disclosure template or a single disclosure event. Template formation is what converts a filing into a durable market instrument.

Group Five — State and Regulatory Response

Nine predictions testing what legislatures, commissions, and utilities do next.

24 — Filing Enters a State Record [Primary] Conditional on a cohort member filing publicly, at least one state legislative, executive, or utility-regulatory record cites information drawn from that filing within twelve months.Confidence: 55–65%, event class. Falsifier: no such citation appears within twelve months. Unactivated if no cohort filing becomes public. Settlement: committee reports, bill reports, agency filings, commission orders, or executive releases.

25 — Peer Proceedings [Secondary] Three or more additional states open large-load tariff or data center standards proceedings citing ratepayer cost allocation. Confidence: 70–80%, event class. Falsifier: fewer than three additional states open such proceedings. Settlement: state commission dockets and executive orders. Deadline: December 31, 2027.

26 — Financial-Assurance Diffusion [Secondary] At least three additional state tariffs, commission orders, or statutes adopt new credit-support, financial-assurance, usage-disclosure, or project-cost protections for very large loads.Confidence: 70–80%, event class. Falsifier: fewer than three such adoptions. Settlement: state commission orders and enacted statutes. Deadline: December 31, 2028.

27 — Texas Codification [Secondary] Texas enacts legislation incorporating at least two of the following, each counted where the enrolled bill contains an operative requirement rather than a study or intent provision: developer-funded grid infrastructure cost responsibility; mandatory electricity or water use disclosure; cooling efficiency or water reuse standards; noise, setback, or community protection requirements; changes to data center tax incentives; or on-site generation or capacity obligations. Confidence: 75–85%, event class. Falsifier: no enrolled Texas bill contains operative requirements in at least two categories. Settlement: Texas Legislature enrolled bills. Deadline: December 31, 2027.

28 — Texas Attrition [Secondary] At least two data center projects, beyond any project already reported as ended or withdrawn as of August 13, 2026, withdraw, are denied, or are indefinitely delayed following the ERCOT audit. Indefinite delay means removal from the interconnection queue, official suspension, or no recorded advancement for two consecutive quarters. Confidence: 55–65%, event class. Falsifier: fewer than two additional projects withdraw, are denied, or meet the delay definition. Settlement: Governor’s office releases, PUCT and ERCOT records. Deadline: June 30, 2027.

29 — Wisconsin Credit Standard [Secondary] The Public Service Commission of Wisconsin does not materially relax the A- credit standard on rehearing, where material relaxation means removing the standard, raising the eligibility threshold above 250 megawatts, or eliminating the financial-assurance requirement. Confidence: 55–65%, event class. Falsifier: the Commission takes any of the three named relaxation actions. Settlement: PSCW docket order. Deadline: June 30, 2027.

30 — Washington Unbundling [Secondary] A Washington bill addressing large-load tariffs or data center disclosure is introduced in the 2027 session without accompanying tax-exemption changes. Confidence: 50–60%, event class. Falsifier: every 2027 data center bill couples tariff or disclosure provisions to tax-preference changes, or no such bill is introduced. Settlement: Washington State Legislature bill records. Deadline: 2027 sine die.

31 — Washington Coupling [Secondary] Conditional on a 2027 Washington data center bill being introduced that couples tariff or disclosure provisions to tax-exemption changes, that bill fails to reach a floor vote in the second chamber.Confidence: 60–70%, event class. Falsifier: such a bill reaches a floor vote in the second chamber. Unactivated if no coupled bill is introduced. Settlement: Washington State Legislature bill records. Deadline: 2027 sine die.

32 — Consumer-Owned Utility Response [Secondary] A Washington consumer-owned utility or public utility district adopts a large-load service policy, contract standard, or tariff equivalent. Confidence: 40–50%, event class. Falsifier: no Washington consumer-owned utility or public utility district adopts such an instrument. Settlement: utility board records. Deadline: December 31, 2027.

Prediction 24 closes the loop the rest of the note argues. A securities filing cited in a state record converts the disclosure claim from prediction into demonstrated transmission between two systems that have never spoken to each other directly.

Group Six — Political Risk Entering the Record

Six predictions running the causal direction the other groups leave open. Groups Two through Five test what filings disclose and what states do next. Group Six tests the return path — whether legislative sessions, moratoria, and organized local opposition reach offering timing, pricing, and analyst judgment.

Most state legislatures convene between January and April 2027, which places the first full session cycle directly inside the window when the cohort expects to price.

33 — Session-Specific Legislative Risk [Secondary] A cohort registration statement or post-listing periodic report names a pending or anticipated state legislative session, bill, or ballot measure as a risk to power cost, siting, or project timing. Confidence: 60–70%, event class. Falsifier: no cohort filing or periodic report names a pending or anticipated legislative or ballot action. Unactivated if no cohort filing becomes public. Settlement: EDGAR. Deadline: December 31, 2027.

34 — Opposition-Attributed Project Impact [Secondary] At least one cohort issuer discloses a project delay, cancellation, downsizing, or cost increase attributable to local opposition, a moratorium, a referendum, or a permit denial.Confidence: 55–65%, event class. Falsifier: no cohort issuer attributes a project impact to any of those causes. Unactivated if no cohort filing or periodic report becomes public. Settlement: EDGAR and issuer communications. Deadline: December 31, 2027.

35 — Session-Window Offering Effect [Secondary] Conditional on a cohort offering pricing between January 1 and June 30, 2027, at least one contemporaneous state legislative or regulatory action is identified as affecting that offering’s timing, size, or pricing. Confidence: 40–50%, event class. Falsifier: no such action is identified in issuer communications, underwriter research, or two or more named financial press reports. Unactivated if no cohort offering prices inside the window. Settlement: issuer communications, published research, and named financial press reporting.

36 — Research Prices Legislative Risk [Primary] At least one published sell-side note on a listed cohort issuer names state legislative, regulatory, or local siting action as a factor in its valuation, rating, or price target. Confidence: 65–75%, event class. Falsifier: no published note names such action as a valuation, rating, or price-target factor. Unactivated if no cohort issuer is publicly listed and covered. Settlement: published equity research. Deadline: December 31, 2028.

37 — Quantified Moratorium Exposure [Secondary] A cohort filing or periodic report quantifies exposure to local moratoria or siting restrictions by naming a count of affected sites, megawatts, or jurisdictions. Confidence: 35–45%, event class. Falsifier: no cohort filing quantifies such exposure. Unactivated if no cohort filing becomes public. Settlement: EDGAR. Deadline: December 31, 2027.

38 — Post-Listing Policy Posture [Primary] Within four quarters of listing, the first listed cohort member publicly commits to a state standards framework, a ratepayer-protection tariff structure, or a community benefit agreement it had not committed to before listing. Confidence: 60–70%, event class. Baseline set at listing: the issuer’s publicly recorded policy commitments as of the pricing date. Falsifier: no commitment beyond that baseline appears within four quarters. Unactivated if no cohort listing completes. Settlement: governor and commission releases, utility filings, local government records, and issuer communications.

Group Six establishes that legislative and local risk reaches filings, research, and post-listing conduct. Group Seven tests whether the risk reaches price.

Group Seven — Effects on Share Price and Offering Execution

Five predictions measuring whether authorization and political risk produce observable effects on valuation, offering execution, and investor practice. Groups Two through Six establish visibility. Group Seven establishes consequence.

Every prediction below settles against market data, prospectuses, or published methodology rather than against language in a filing.

39 — Adverse-Action Price Reaction [Primary] Following public disclosure of a named adverse state or local authorization event affecting a material campus of a listed cohort issuer, that issuer’s shares underperform the Pacer Benchmark Data & Infrastructure Real Estate ETF by at least three percentage points across the five trading days after disclosure. Confidence: 50–60%, event class. Clean-event condition: the five-trading-day window carries no earnings release, financing announcement, major customer announcement, guidance change, or index reconstitution affecting the issuer. Falsifier: underperformance of less than three percentage points across the window on the first qualifying clean event. Unactivated if no qualifying clean event occurs. Settlement: closing prices for the issuer and the named benchmark. Deadline: December 31, 2028.

40 — Sustained Multiple Spread [Secondary] Two listed cohort issuers sustain a valuation-multiple gap of at least twenty percent, measured on enterprise value to EBITDA or price to funds from operations, across a full calendar quarter.Confidence: 60–70%, event class. Prediction 40 measures the spread only and stays outside the thesis score unless the Prediction 19 attribution rule also settles correct, since leverage, growth, customer mix, geography, and margin structure each produce multiple gaps independent of authorization. Falsifier: no such gap persists across a full quarter. Unactivated if fewer than two cohort issuers are publicly listed for a full quarter. Settlement: reported financials and closing prices. Deadline: December 31, 2028.

41 — Later Listings Price Lower [Secondary] Conditional on Prediction 11 settling correct and at least two cohort listings pricing, the second listing prices at a lower implied valuation per megawatt of disclosed operating capacity than the first. Confidence: 55–65%, event class. Measurement rule: operating capacity and authorized-but-unenergized capacity are compared separately, so a large speculative pipeline cannot improve the ratio. Falsifier: the second listing prices at or above the first on operating capacity. Unactivated if fewer than two cohort listings price, or if operating capacity is not separately disclosed. Settlement: final prospectuses and pricing announcements. Deadline: December 31, 2028.

42 — Allocator Methodology Response [Secondary] At least one institutional investor, index provider, or credit rating agency publishes methodology, criteria, or formal commentary treating authorization status, interconnection position, or power deliverability as a distinct assessment factor for data center issuers. Confidence: 50–60%, event class. Qualifying documents must be newly published or materially revised after August 13, 2026. Falsifier: no such new or revised methodology, criteria, or commentary appears. Settlement: published methodology and criteria documents. Deadline: December 31, 2028.

43 — Offering Execution Effect [Secondary] At least one cohort offering is postponed, downsized, or repriced with authorization, power availability, or regulatory risk identified as a contributing cause. Confidence: 40–50%, event class. Falsifier: no cohort offering is postponed, downsized, or repriced on those grounds. Identification requires issuer or underwriter statement, or two or more named financial press reports. Unactivated if no cohort offering reaches marketing. Settlement: SEC filings, issuer communications, and named financial press reporting.

Prediction 39 supplies the sharpest single test in the register. A measurable share-price reaction to a state or local authorization event, on a clean event window, provides the strongest observable evidence that public markets price authorization risk continuously.

How the Sequence Loops Back to the Statehouse

Seven groups trace one loop, and each link carries its own registered forecast.

1. States write explicit terms. Wisconsin set a fifteen-year commitment and a collateral standard. Texas paused an interconnection queue pending audit. Predictions 25, 26, 27, and 29.

2. Filings disclose the exposure. Materiality drives capacity distinctions and named jurisdictions into the registration statement, and may surface specific authorization instruments. Predictions 8, 9, 11, 12, 13, 33, and 37.

3. Regulators and research read the filings. SEC staff comments; analysts build authorization into frameworks and price targets. Predictions 16, 19, 36, and 42.

4. Price responds. Offerings discount development pipelines, multiples separate, adverse events move shares. Predictions 17, 18, 39, 40, 41, and 43.

5. Issuers change conduct. A listed developer facing priced political risk makes public commitments it avoided while private. Prediction 38.

6. States read the filings and the conduct. Legislative, executive, and regulatory records cite what the filings disclosed. Prediction 24.

Step six returns to step one. Authorization terms written in a statehouse arrive back at that statehouse eighteen months later carrying a price attached by public markets, and the register above dates every link in the sequence.

Thesis Test

Two separable claims carry the authorization-multiple thesis, and each fails on its own terms.

Disclosure test. The first cohort registration statement names a state or regulatory regime and quantifies materially different capacity states. Generic boilerplate defeats the claim. Naming a specific instrument, tested at Prediction 9, counts as stronger confirmation rather than a necessary condition.

Valuation test. Published research applies an authorization-linked valuation differential under the Prediction 19 matching rule, or a qualifying clean authorization event produces the relative-price effect specified in Prediction 39. Differentiated disclosure without either result defeats the claim.

Public documents settle both tests on stated deadlines. Visibility and pricing are separate achievements, and the register scores them separately.

IX. Scope and Limitations

Two limits bound the analysis, and both concern how much the August 13 report can carry.

Reuters described preliminary deliberations rather than a launched process, and sources disagreed materially on offering size — one figure near $10 billion, another near $5 billion. Disagreement of that magnitude indicates a transaction without settled shape.

Reuters also did not state whether the $100 billion figure represents equity value or enterprise value. Reuters expressly described Switch’s comparable figure as including debt, which makes the omission meaningful. Equinix currently carries a market capitalization near $103 billion and Digital Realty near $73 billion, so the reading chosen determines whether Vantage would enter at, above, or below the established public operators. The figure functions as a sponsor-side anchor rather than a discovered price.

MindCast anchors the analysis to the four-platform cohort rather than to Vantage alone. Should Vantage never file, the thesis settles against whichever platform files first.

Source Notes

The principal factual claims above trace to the following public records.

Texas audit directive, August 3, 2026 — Office of the Governor of Texas.

Vantage compliance announcement, August 12, 2026 — Office of the Governor of Texas.

Vantage transaction deliberations, August 13, 2026 — Reuters.

Wisconsin Very Large Customer tariff order, April 24, 2026 — Public Service Commission of Wisconsin.

We Energies rehearing request on credit standards, June 2026 — Wisconsin Watch.

Switch confidential filing, August 7, 2026 — Bloomberg, reported by Reuters.

Cohort listing plans — Data Center Dynamics.

Frontier campus, Shackelford County, Texas — Vantage Data Centers.

Lighthouse campus, Port Washington, Wisconsin — Vantage Data Centers, OpenAI, and Oracle.

Quincy campus completion, Grant County, Washington — Vantage Data Centers.

Washington House Bill 2515 outcome — Davis Wright Tremaine.

CyrusOne data center locations, including Quincy, Washington — CyrusOne.

Switch Pennsylvania campus and cost commitment, April 2026 — Las Vegas Review-Journal.

Foundational Works

AI Data Center Credit Risk supplies the instruments applied here — Authorization Credit Exposure, Financeable Authorization Coverage, Capital Irreversibility, and the Financing Inversion hypothesis.

The Data Center Authorization Price establishes the fifty-state baseline: every project pays an authorization price, and certainty can be worth more than nominal leniency.

The 50-State Authorization Price Atlas supplies the jurisdiction-level conditions behind any national portfolio, and explains why no developer’s capacity deserves a single undifferentiated risk score.

The Model Data Center Authorization Code supplies the measurement scale against which disclosed authorization risk can be graded.

Data Center Investor Confidence establishes that developers pay a second time when capital providers price the credibility of delivery.

Three Competing Governance Equilibria sets out the state-policy regimes that would alter the value of a geographically distributed portfolio after any listing.

The MindCast Data Center One-Year Review records what MindCast identified about authorization, cost allocation, flexibility, and energization risk before the current valuation test.

Related works: AI Data Center Developers Going Public in America Must Disclose Foreign Ownership and Control | AI Data Center Credit Risk — Permitting, Curtailment, and the Cost of Capital · The MindCast AI Data Center Record (August 2026) · The Authorization Market: Standardized Bargaining, Rationed Power, and the Competition to Build America’s AI Infrastructure (2026) · The Data Center Authorization Price: A 50-State Baseline (2026) · Capital Is the New Computing: Financing the Next Era of AI Infrastructure, 2025–2045 (November 2025) · Power Brokers & Digital Real Estate: How CRE Firms Are Building the AI Infrastructure Backbone (November 2025)

Critical references: The Model AI Infrastructure Authorization Code · The Data Center Authorization Market: A 50-State Regulatory Atlas

Executive Summary

Washington sets an authorization price on advanced compute infrastructure built anywhere in the world.

A campus in Riyadh can hold every domestic permit, complete construction, energize its substation, and still not run. Advanced compute reaches it only under a United States export license, and no Saudi instrument governs the decision. Host authorization determines whether a building goes up. A second authorization, set thousands of miles away, determines whether the building does the work it was built for.

Two authorization prices govern advanced compute infrastructure, and the two answer different questions. The site authorization price covers permission to site, connect, power, and operate — statute, tariff, commission order, and local ordinance, graded across fifty American states and ten foreign jurisdictions in the companion publication. The strategic sovereign authorization price covers controls over technology access, ownership, capital movement, and strategic connectivity: export licensing, investment screening, outbound investment rules, and submarine cable landing approval.

Site Authorization Price + Strategic Sovereign Authorization Price = Total Authorization Exposure.

Separating the two functionally rather than geographically matters, because a single government often holds both. The United States sets site authorization in Ohio and strategic authorization over a project in Riyadh, and the two decisions run through different agencies on different logic. Capital pricing only the first term is pricing part of the risk.

The strategic price falls unevenly, and the unevenness maps cleanly. A Canadian pension fund investing in Ohio faces one instrument. A Gulf sovereign investing in the same asset faces several, with two material ones — investment screening and export licensing for downstream access — originating in a single American strategic posture and moving together.

The same Gulf allocator funding a campus in Vietnam adds Vietnamese ownership rules, a currency and repatriation regime, and its own government’s outbound rules. Decomposing the strategic price by instrument produces the Sovereign Exposure Stack, which identifies routing incentives that expected-return analysis alone misses.

Geopolitical pressure changes how capital is structured before it changes how much capital moves. Conventional analysis expects tension to reduce cross-border flows. The mechanism runs the other way first. When access to American technology becomes less certain, a sovereign allocator has more reason to hold American assets, not less, because ownership sits closer to the technology. Domestic buildout accelerates at the same time as a hedge. Both legs speed up, and volume falls only after structure has exhausted its options.

Four structures carry that adaptation, and one of them changes what states can do. Debt in place of equity, platform intermediation, dedicated on-site generation, and powered-shell underwriting each reduce a foreign allocator’s exposure to host regulatory action. Three affect visibility. The fourth affects power.

Bring-your-own-generation moves the binding authorization instrument away from the standard large-load queue, and the queue is the principal rationing instrument in the grid-adequacy corrections examined here. Alberta rations with it. Quebec allocates with it. Ireland’s four-year Dublin closure ran entirely through it.

Leverage does not vanish when a project leaves the queue. A project supplying dedicated generation still touches transmission, standby service, generation siting, and emissions approval, so leverage migrates to instruments held by different bodies on different timetables. Ireland and Alberta both adopted bring-your-own-generation as their correction, transferring the adequacy problem to the applicant and dispersing their own leverage across instruments they had not been using to ration.

Powered-shell underwriting works in the same direction. An asset engineered to yield a real estate return without advanced chips lowers the cost of delayed compute eligibility, which changes how an allocator responds to licensing uncertainty. Utilization, tenant demand, and valuation all differ sharply from an operating AI facility, so the structure buys time rather than immunity.

The same instrument means opposite things on opposite sides of the study. Where a grid delivers, dedicated generation bypasses queue-based rationing and shifts leverage to instruments held by other bodies. Where a grid cannot supply underwritable firm power on the required timetable — Brazil, and parts of India and Vietnam — dedicated or contractually firm generation becomes an entry condition rather than a bypass. One instrument, two meanings, decided by whether the grid can deliver.

Five Predictions Carry the Argument

Naming the deeper mechanism: authorization-instrument substitution. The companion publication showed capital responding to a higher authorization price by changing where an asset sits. The present paper shows capital responding by changing what the asset legally and economically is — equity to debt, direct to intermediated, grid-served to self-generated, compute-dependent to powered shell. Each change moves which government instrument still binds.

Jurisdictional substitution and structural substitution are the two substitution margins available in response to a price increase, and only the first shows up in siting data. Cancellation remains the third response, and the scenario weights in Part VIII carry it.

The conclusion for American jurisdictions is concrete. Geopolitical pressure should produce structural substitution before it produces large-scale geographic substitution, because changing a deal structure is faster and cheaper than moving a site. A state that writes a bring-your-own-generation pathway without asking where leverage migrates has changed the negotiation without noticing.

What Each Stakeholder Gets

Parts I through III build the instrument. Parts IV and V examine each leg of the flow. Parts VI and VII set out what follows for capital and for policy. Part VIII releases the prediction register.

I. Two Authorization Prices

Advanced compute infrastructure requires two permissions answering two different questions, and most analysis prices only one.

Site authorization answers whether the facility may exist and operate. Statute sets the conditions, tariff sets the cost allocation, a commission order sets the terms of service, and a local ordinance sets whether the project may occupy the ground. The companion publication defines the authorization spread on exactly those four components, conditional on power delivery timing, technology access, and tenant credit quality.

Strategic sovereign authorization answers whether the facility may do the work it was built for, under the ownership contemplated, connected to the markets intended. Advanced compute moves under export license. Foreign ownership of infrastructure moves under investment screening. Capital moving across borders increasingly moves under outbound rules. Submarine cables reaching a landing point move under separate security review.

Ordinary enterprise and colocation facilities carry little strategic exposure. Facilities built for frontier compute carry a great deal, which is why the distinction matters now and did not a decade ago.

A host government frequently sets the first price and cannot set the second. Riyadh can grant every permit a campus requires and cannot license a single accelerator into it. Hanoi can approve a project in weeks and cannot guarantee the silicon that makes the project economic.

The asymmetry is the subject of this paper. A jurisdiction competing on authorization terms competes on a variable it controls, against a constraint it does not, and capital prices both together.

II. The Sovereign Exposure Stack

Decomposing the strategic sovereign price by instrument produces an exposure topology. For any allocator and any destination, the Sovereign Exposure Stack lists the distinct strategic instruments that can materially alter the position — controls over technology access, ownership, capital movement, and connectivity — together with the degree to which those instruments move as a bloc.

Site authorization sits outside the stack. Ordinary permitting, tariff, and ordinance risk belongs to the companion publication’s spread, and the two combine only at the level of total exposure.

Depth matters, and correlation can dominate depth. Instruments originating in one government under one strategic posture move together, so an allocator facing four correlated instruments carries a different risk than an allocator facing four independent ones.

The stack maps exposure rather than pricing it. Three instruments are not automatically riskier than one, because a single instrument with wide discretion and severe consequence can dominate several narrow ones. A fully specified version would weight depth against correlation, decision severity, and reversibility, and no such weighting has been validated against outcomes.

Stack topology surfaces behavior that return alone leaves unexplained. The model implies that an allocator facing a single instrument optimizes for yield, while an allocator facing three correlated instruments optimizes for structure first and yield second — which would explain why Gulf capital and Canadian capital behave differently in the same asset at the same price. P3 tests the implication directly.

Construction requires published rules and no proprietary data, so any allocator can build the stack for its own book before committing.

III. Five Vectors

Geopolitical events reach data center capital through five channels. Each operates differently, and treating them as one undifferentiated risk produces the wrong hedge.

Export licensing and chip eligibility. Advanced accelerators reach a destination only under license, and eligibility varies by country and by end user. A campus is a real estate asset until the silicon arrives, at which point the asset becomes compute. Licensing therefore governs the conversion, not the construction.

Investment screening, inbound and outbound. Foreign acquisition of American infrastructure draws review that varies by structure and by acquirer identity. Sovereign vehicles draw the most scrutiny in every structure. Outbound rules, newer and expanding, govern American capital moving into foreign technology assets.

Submarine cables and landing rights. Cable geography determines which markets a facility can serve, and landing approval is a sovereign decision subject to security review. Cable damage in contested waters closes routes outright. A substitution corridor requires physical connectivity as a precondition, and cable geopolitics is the only instrument in this set that can close a corridor completely rather than raise its price.

Upstream concentration. Advanced packaging, high-bandwidth memory, large power transformers, grain-oriented electrical steel, and critical minerals each concentrate in few suppliers and few jurisdictions. Concentration converts an ordinary procurement question into a geopolitical one, and transformer lead times already govern energization schedules independently of any authorization decision.

Sovereign AI as industrial policy. National programs across the European Union, India, Japan, and the Gulf direct capital toward domestic capability for reasons that are not commercial. Programs of that kind move capital against price signals, which makes them a genuine competitive factor and an unreliable one.

Two classes of constraint sit inside the five vectors, and the distinction governs how an allocator should hedge.

Feasibility constraints create binary breaks. Export licensing and cable landing access can each stop a project that has satisfied every host requirement, and neither is hedgeable through pricing.

Cost and timing constraints shift expected value. Investment screening, upstream concentration, and sovereign industrial policy change what a position costs, when it delivers, and against whom it competes, and each can be priced.

Confusing the two produces the wrong instrument. A feasibility break requires optionality — structure, staging, alternative use. A cost shift requires a spread.

Subscribe now

Working With MindCast

MindCast AI is a predictive law and behavioral economics firm in Bellevue, Washington. The practice models institutions and firms as decision systems — Cognitive Digital Twins carrying objective functions, constraint stacks, and failure modes — then runs them forward to produce time-bound predictions with named falsifiers and settlement sources. Every forecast MindCast publishes can be graded against a public record by a stated date.

Two verticals now run on the same instruments. Authorization intelligence grades statute, tariff, commission order, and local ordinance across all fifty states and ten foreign jurisdictions, and supplies the AI Infrastructure Authorization Series and the 50-State Authorization Price Atlas. Geopolitical exposure intelligence, introduced in the present paper, maps the controls over technology access, ownership, capital movement, and connectivity that condition an asset after the host has already approved it.

Published research grades jurisdictions and mechanisms in general terms. A specific position requires specific work.

MindCast produces market-specific and counterparty-specific analysis on request:

• Sovereign Exposure Stack for a named allocator across a defined destination set

• Authorization-spread assessment for a cross-border pipeline or a single site

• Jurisdictional authorization profile for any American state, county, or foreign jurisdiction in the study

• Structural substitution review — which instrument becomes binding under a proposed deal structure, and which body holds it

• Multi-jurisdiction comparison for a site-selection or capital-allocation decision

Government affairs teams, investment committees, developers, utilities, and legislative staff each receive the same underlying instruments applied to their own facts.

Contact MindCast AI at mcai@mindcast-ai.com.

Visit the MindCast Corporate Site

IV. Capital Into the United States

Foreign capital entering American data center assets faces the host’s own apparatus, and the companion publication names the four exposures: cost-allocation reversal, local veto latency, queue-position value decay, and regulatory exposure correlated by capital origin.

Structure is the primary response, and four structures recur.

Debt in place of equity. A senior secured position absorbs a valuation shock differently from an equity position. A commission order reallocating transmission cost after commitment reduces equity value directly, while a lender retains claims on land, improvements, and routing.

Platform intermediation. Capital placed alongside an established domestic manager reaches assets through a counterparty that already operates in the relevant regulatory forums. Intermediation changes which entity appears before a utility commission or a county board.

Dedicated on-site generation. A project supplying its own generation does not depend on the utility’s standard interconnection queue for firm power.

Powered-shell underwriting. A facility engineered to return on real estate fundamentals — shell, substation, cooling, fiber — holds value if advanced compute arrives late or not at all.

Three of the four change what a state can see. The third changes what a state can do, and Part VII takes up the consequence.

Structural adaptation of this kind is ordinary optimization rather than evasion. Different structures carry different review treatment under published rules, and allocators respond to published rules. Naming the response as ordinary keeps the analysis usable by both sides of a negotiating table.

V. Capital Into the Ten Foreign Jurisdictions

Capital moving into the United Arab Emirates, Saudi Arabia, Singapore, Japan, Canada, Ireland, France, India, Vietnam, or Brazil carries three risks the inbound American leg does not.

Chip eligibility is set by a third party. An allocator funding a campus in Hanoi or Hyderabad faces a licensing question that neither Vietnam nor India can control. Host authorization policy, however well designed, does not address the dominant risk. A host government can shorten its approval timeline to weeks and change nothing about whether the facility receives accelerators.

Permission itself can be withdrawn. American commissions reallocate cost and rarely revoke permission. Chile’s courts unwound a granted permit over groundwater, and any jurisdiction whose judicial or environmental review retains reversal power quotes a certainty it cannot guarantee.

Currency, capital controls, and repatriation price into Brazil, India, and Vietnam and price into none of the American exposure. A return denominated in a currency subject to control is a different asset from the same return in dollars.

Against those risks, dedicated generation reverses meaning. Where a grid cannot supply underwritable firm power on the required timetable — the condition in Brazil, and in parts of India and Vietnam — dedicated or contractually firm generation becomes an entry condition rather than a regulatory bypass. Bring-your-own-generation routes around an instrument in a jurisdiction whose grid delivers and substitutes for the grid in a jurisdiction whose grid does not.

The reversal explains a pattern that otherwise reads as inconsistency. An allocator can adopt the same structure in Alberta and in Bahia for opposite reasons — in one case to move outside a rationing instrument, in the other to substitute for a grid that will not deliver.

VI. What the Combination Produces

Two legs and five vectors interact, and three consequences follow that neither leg produces alone.

Chip eligibility becomes an explicit underwriting input. Accelerator availability is often modeled as a supply or procurement input. Strategic eligibility requires separate treatment, because eligibility is a sovereign determination varying by destination and by end user, and no standard instrument exists for measuring it.

Capital adapts through structure before it adapts through geography. Separating what is observed from what the model implies matters here, because the two carry different weight.

Observed: deal structures differ measurably in exposure and in review treatment under published rules.

Implied: rising strategic pressure should therefore produce structural adaptation — debt, intermediation, dedicated generation, powered shell — before it produces volume contraction, because changing a transaction is faster and cheaper than moving or cancelling a project.

Testable: whether deal-form migration precedes any decline in capital flow. Part VIII scores the sequence as P1 and P8.

Host assurances on technology access face an incentive problem worth testing.

Observed: chip eligibility is a sovereign determination of the exporting state, and host governments hold no instrument that binds it.

Implied: a jurisdiction competing for capacity therefore has an incentive to assure investors on access it cannot secure, and allocators have reason to discount such assurances.

Testable: whether host assurances systematically precede licensing outcomes the host does not control.

Together the three consequences point one way. Geopolitical pressure reorganizes the form capital takes before it reduces the amount, and form is exactly what host authorization law regulates least well.

VII. What Follows for American Jurisdictions

Geopolitical pressure reaches American jurisdictions as a structural problem before it reaches them as a geographic one. Changing a deal structure takes months. Moving a site takes years, and for latency-bound workloads may be infeasible at any price.

The interconnection queue is the principal rationing instrument in the grid-adequacy corrections examined here.Alberta rationed 19,565 megawatts of requests against a 1,200 megawatt limit using nothing but queue position. Quebec restricted allocation the same way. Ireland’s four-year Dublin closure operated entirely through connection policy, with no statute and no vote.

A dedicated-generation pathway moves leverage rather than destroying it. Ireland and Alberta both adopted bring-your-own-generation as their correction, transferring the adequacy problem to the applicant. A project outside the standard large-load queue cannot be rationed by the body running that queue, and the same project still requires transmission service, standby arrangements, generation siting approval, emissions permitting, fuel supply, and water.

Leverage therefore migrates to instruments held by different bodies on different timetables, and the receiving bodies are frequently the ones with the least practice using them for this purpose.

Dedicated generation solves a real adequacy problem and solves it well. The question a jurisdiction should ask on adopting the pathway is narrower and answerable: which instrument becomes binding once the queue no longer is, and does the body holding it know that it now sets the price?

Powered-shell underwriting shifts the federal lever in the same way. An asset returning on real estate fundamentals lowers the cost of a licensing delay, so a project proceeds on shell economics and the technology question resolves on a later timetable under different pressure.

Three propositions follow for state and county decision-makers.

Count what you still control after each exemption. An authorization regime is only as strong as the instrument that survives the most common structure used to route around it, and the surviving instrument is usually held by a different agency than the one that wrote the exemption.

Visibility is a policy variable. Debt and intermediation are lawful and ordinary, and both reduce what a commission or a council knows about the party behind a project. A disclosure requirement attached to authorization rather than to ownership survives structural adaptation.

Foreign competition remains the smaller and slower risk. The companion publication holds that latency-bound American serving capacity does not migrate at material scale, and nothing in the geopolitical layer changes the finding. Training and batch workloads are mobile and do move; the capacity serving American users is not, and structural adaptation arrives long before either.

VIII. The Foresight Register

Twenty-one predictions follow from a thirty-one CDT register run against a common August 12, 2026 observation record. Every entry carries a deadline, a falsifier, a named settlement source, and a confidence band with a class label. Event-class entries resolve against an observable occurrence. Interpretive-class entries express confidence in a structural read and never pool with event probabilities.

VIII.1 Primary Register

Eight primary predictions test the mechanisms this paper introduces. Each would fail if the underlying theory were wrong.

Settlement resolves in three states, and the third is narrow by design. Private structures disclose late, so P1, P3, and P8 carry real observation risk. Handling that risk by exempting non-observation from disconfirmation would produce an asymmetric rule in which evidence can confirm a prediction and absence can never falsify it.

HIT — the predicted event is observed under the registered definition. MISS — adequate settlement data exist for the window and the predicted event is not observed. UNRESOLVED — the named settlement source was unavailable or affirmatively incapable of resolving the question.

Ordinary non-observation counts as a MISS. UNRESOLVED requires an affirmative showing that the source failed, not merely that the event went unrecorded.

Three primary entries resolve against a quantitative threshold rather than a binary occurrence: P2, P3, and P6, each carrying its threshold in the entry itself. P2 additionally carries a distribution on the increase in contested siting or air proceedings — P10 20% · P50 45% · P90 75%. Quantiles are model-implied scenario values rather than calibrated statistical estimates.

VIII.2 Secondary Register

Thirteen secondary predictions test individual instruments and jurisdictions. Each resolves against a single named source and on a shorter clock, which makes the secondary entries the early-warning layer for the primary ones. Falsifiers and settlement sources for the secondary entries are maintained in the technical record and available on request; the table below carries the claim, the deadline, the class, and the band.

Four entries resolve within roughly four quarters — S3, S7, S11, and S12 — and each tests a different instrument. Early movement across all four would corroborate the primary register well ahead of its own deadlines.

VIII.3 Scenario Weights and Decisive Uncertainty

Three routes carry the modal analysis, weighted by share of surviving model branches rather than by calibrated market probability. Structural adaptation without geographic relocation runs 55–65%. Structural adaptation followed by partial geographic rebalancing runs 25–35%. A strategic feasibility break admitting no structural workaround runs 10–20%.

Timing and magnitude of the next material licensing or screening action remain exogenous. Neither is forecastable from the routed function set, and P1, P3, S10, S11, and S12 all condition on a trigger this analysis does not predict.

The response window is a different question, and the register does forecast it. Once a trigger arrives, P1 sets the interval between structural change and any change in committed capital at two quarters or more. P7 sets the interval between a structure becoming public and the receiving body tightening conditions. P8 sets the order in which adaptation occurs.

Together the three describe how long an allocator has to act and how long a regulator takes to respond — the part of the sequence that repeats regardless of which trigger fires.

Predicting the day a restriction lands is a different exercise from predicting what happens in the four quarters afterward, and only the second is a structural question.

IX. Method and Scope

Findings and predictions above derive from a thirty-one CDT register: the twenty-three jurisdictional and firm-side CDTs built for the companion publication, plus eight covering sovereign instrument-holders, allocator classes, and the bodies that receive migrated leverage. Two independent executions ran the register against a common observation record.

Three limits bound the analysis.

Rules move faster than publication. Export-control and screening rules change on their own schedule, so every characterization here reflects published rules as of August 12, 2026.

Private structures disclose late. Structural adaptation is observable only where disclosure requires it, and private credit and unlisted vehicles report incompletely. The aggregate shift toward debt and intermediation is therefore directionally supported rather than precisely measured, and Part VIII names the entries carrying the resulting false-negative bias.

The Sovereign Exposure Stack lists instruments rather than weighting them, because no weighting has been validated against outcomes.

The three limits constrain measurement precision and settlement observability. None alters the causal mechanisms the model generates, which rest on published rules and on completed corrections settled against public record.

See MindCast AI Data Center Live-Fire Mission

Executive Summary

Wall Street began pricing community data center opposition in July-August 2026. MindCast published the mechanism nine months earlier.

On August 10, Reuters reported that banks and asset managers financing US data centers now weigh local political resistance directly in credit assessment. Karen Fang, who runs infrastructure and sustainable finance at Bank of America, described two underwriting tests: project readiness and credit quality. Fang defined readiness as the permits a project requires plus the support of the people who will live around it. Senior bankers said they now favor projects in states more welcoming to data centers.

MindCast forecast the shift in November 2025 and dated the window. Four calls from the investor and capital work have now settled against outside evidence.

Community trust becomes the decisive site-selection filter — Ship the Workload, Not the Power, September 2025. Gallup reported in May 2026 that roughly seven in ten Americans oppose a data center in their area. Blackstone-owned QTS terminated the Prince William Digital Gateway after sustained local opposition.

Lenders will underwrite energization timing as rigorously as tenant creditworthiness — Power Brokers & Digital Real Estate: How CRE Firms Are Building the AI Infrastructure Backbone, November 2025. Reuters confirmed the practice on August 10. CyrusOne’s $9.7 billion facility releases construction capital only where permits and leases are complete.

Community standing enters financing terms between the third quarter of 2026 and the second quarter of 2027 — Building CRE Coherence for AI Infrastructure: The Integration Imperative, November 2025. The August 10 reportinglanded in the first quarter of that window.

Transformer and transmission scarcity are real; nationwide grid collapse is not — The AI Infrastructure Energy Opportunity Landscape, March 2026. McKinsey confirmed both readings within four months, reporting equipment lead times more than doubled since 2019 and grid connection waits exceeding four years.

The mechanisms settled. The magnitudes did not, and Section II grades both.

The publication carries two claims, and they hold at different strengths. Separating them is the point.

Supported. Authorization risk has become capital-market risk. Uncertainty about permission now changes the timing, probability, and recoverability of the cash flows supporting capital, and lenders price it.

Hypothesised, and testable. Capital structure may also determine how much a state can extract from a developer. A company building from operating cash flow can walk away from a hostile jurisdiction and return in a later session. A company paying interest on drawn construction debt, holding non-cancelable equipment orders, and facing a contractual delivery date cannot.

Capital structure increases a firm’s capability to build, while sufficiently irreversible committed capital decreases its ability to wait.

MindCast names the mechanism the Financing Inversion. Prediction P14 settles it.

What This Publication Adds

A translation layer. States negotiate in tariffs, collateral, minimum-take terms, and months to energization. Credit markets negotiate in coverage ratios, duration, loss severity, and basis points. The conversion between them is what this publication attempts.

Two measurement instruments. Financeable Authorization Coverage measures how much of a project’s committed capacity has cleared every gate it needs. Capital Irreversibility measures how much of its money can no longer move. Together the pair identifies which projects are hold-up exposed — the same question a state asks when deciding how hard to press.

Four live exposures from the last six weeks. Curtailment obligations arriving in June 2027, now carrying an explicit escape price through PJM's bilateral capacity matching. Ownership opacity that sponsors create in the financing room and pay for in the hearing room. A July 29 SEC staff letter that lifted mandatory risk retention from a class of data center securitizations in the same month authorization risk entered credit analysis. And a $500 billion GPU-backed financing platform announced August 10, placing a second collateral assumption — residual hardware value — underneath the same structures.

Evidence that the mechanism already binds. Blackstone-owned QTS terminated the Prince William Digital Gateway after sustained opposition, never approaching lenders for bank financing. CyrusOne’s $9.7 billion facility releases construction capital only where permits and leases are complete. Blue Owl declined a $10 billion Oracle-linked financing in December 2025, weighing Michigan local politics among its reasons.

Prediction Highlights

MindCast generated the predictions below by running the analysis as a MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation across nine behavioral twins — a state government, a grid operator, a county, a community coalition, three developer archetypes distinguished by capital architecture, a bank lender, and a composite creditor. Section XVI states the method; Section XVII carries the full register with settlement sources and falsifiers.

Highest-conviction event predictions.

A top-ten 2026 US project is publicly cancelled or indefinitely paused with permitting or opposition among the stated reasons, by December 31, 2027 — 75–85%.

Data center securitization issuance in 2026 exceeds the 2025 total, by February 28, 2027 — 75–85%.

FERC approves PJM’s Reliability Backstop Procurement in substantially filed form, by December 31, 2026 — 70–80%.

Three or more financings originated after August 11, 2026 disclose permit-to-draw conditions, by June 30, 2027 — 65–75%.

Regulatory concessions get cited by sponsors as evidence of reduced execution risk, by December 31, 2027 — 65–75%.

The instrument test. Projects with less than half their committed capacity fully authorized, and more than sixty percent of their capital already immovable, experience cancellation, delay, or repricing at a higher rate than every other project type — 60–70%. A failure retires both instruments.

Highest-value structural read. Capital concentrates in jurisdictions publishing bounded terms over jurisdictions offering low nominal burden with unresolved rules — 70–80% interpretive.

Provenance. MindCast executed the simulation through an independent Claude implementation of the MP CDT FS runtime. Section XVI states the method and the limits.

A new buyer. Pension funds, insurers, and private credit now hold direct exposure to county zoning boards. None sits at the state bargaining table. None has an instrument for the risk.

What Each Allocator Takes From This Publication

Institutional credit — banks, private credit, infrastructure debt, investment-grade holders. Covenant architecture, rating methodology, curtailment exposure, draw conditions, and sponsor retention after the July 29 staff letter. Predictions P1 through P6 settle on documents these desks already read.

Private equity and infrastructure funds. Irreversibility, optionality, authorization maturity, and project sequencing. Financeable Authorization Coverage and Capital Irreversibility sort a pipeline into four exposure states before capital commits.

Hedge funds and public-equity investors. Capital expenditure exposure, project cancellations, disclosure changes, and the geographic migration of capital. The quarterly settlement cadence in Section XIV matches a public-markets holding period.

Family offices and accredited investors. A usable rule for separating announced infrastructure from financeable infrastructure. Announcement megawatts and authorized megawatts are different assets, and only one of them earns.

Foreign investors and sovereign wealth funds. The sharpest version of the opacity problem. Ownership identity is itself an opposition trigger, so the disclosure remedy that lowers a domestic sponsor’s capital price can raise a foreign sponsor’s authorization price. Section XIII sets out the asymmetry.

Governments deploying capital alongside policy. Extraction has a ceiling. Section VIII names the three boundaries a jurisdiction crosses before it starts raising its own cost of capital.

The publication runs in sequence: what the price is, how it reaches capital, what changed in the last six weeks, how to measure it, what to do, and what MindCast expects to happen. A reader with fifteen minutes should read Sections III, V, and XVII.

I. Two Research Streams and the Gap Between Them

MindCast has worked the AI infrastructure problem from two directions since August 2025. The streams have never been joined, and the gap between them is where this publication operates.

The first stream asked what states charge for permission. Beginning with The Bottleneck Hierarchy in U.S. AI Data Centers in August 2025, MindCast established that electricity, network topology, and cooling set the physical limits on AI infrastructure. The Authorization Market then defined permission to build and operate as an economic good carrying a price, and The Data Center Authorization Price: A 50-State Baseline measured that price in every state on one instrument. Nothing in that work required a reader to know anything about credit markets.

The second stream asked what institutional behavior earns in the credit market. Capital Is the New Computingestablished that infrastructure requires financing durable enough to match a twenty-year asset life. Power Brokers & Digital Real Estate and Building CRE Coherence for AI Infrastructure, both November 2025, modeled twelve platforms and found that developers whose behavior stays consistent under pressure deliver on schedule while inconsistent ones run roughly fourteen months late and twenty-two percent over budget. Nothing in that work required a reader to know anything about state legislatures.

Neither stream asked the joining question: what happens to the authorization price once it enters a project’s capital structure, and how does that structure change the firm’s position at the state bargaining table?

Appendices B and C list both bodies of work for readers who want the underlying studies. No claim in this publication requires them.

The gap matters because an authorization requirement carries no single economic value.

A six-month delay means one thing to a project whose major capital remains undrawn. The same delay means something else entirely to a project carrying billions in construction financing.

A fifteen-year minimum utility commitment means one thing to an investment-grade hyperscaler serving diversified demand. The same commitment means something else to a single-purpose vehicle depending on one tenant, one campus, one energization date, and one refinancing path.

The law is identical in each pair. The financial consequence is not. Closing that gap is the work of the sections below.

II. The November 2025 Forecast and What Settled in August 2026

MindCast grades its own record before extending it. Four dated claims from the Capital stream reached settlement this month, and the grade is mixed by design.

November 2, 2025 — Capital Is the New Computing: Financing the Next Era of AI Infrastructure, 2025–2045. MindCast named the interval between funding commitment and energization as the emerging efficiency benchmark and called it liquidity latency. Projects with documented community engagement, the study found, close financing 40 to 60 percent faster than projects treating stakeholder relations as an afterthought.

November 3, 2025 — Power Brokers & Digital Real Estate: How CRE Firms Are Building the AI Infrastructure Backbone. The study stated that no project reaches financial close without proof of firm power, and that lenders had begun underwriting energization timing as rigorously as tenant creditworthiness. Projects with strong community standing showed a 15 to 20 percent permitting advantage. Phoenix water backlash extended timelines nine months; proactive transparency in the Carolinas halved them.

November 13, 2025 — Building CRE Coherence for AI Infrastructure: The Integration Imperative. The study forecast that by the fourth quarter of 2026, lenders would fold execution-consistency measurement into a material share of infrastructure deal covenants, and that projects without a verified track record would face 50 to 75 basis point penalty pricing. The forecast placed the shift in a window running from the third quarter of 2026 through the second quarter of 2027.

August 10, 2026 — settlement. Reuters reported that banks now weigh permitting readiness and community support directly in project credit assessment.

Three independent measurements explain why lenders moved. Data Center Watch counted at least 75 projects worth roughly $130 billion facing local opposition in the first quarter of 2026 alone, with opposition groups rising from 396 to 833 across 49 states. Carbon Direct found 46 projects worth $170 billion delayed or cancelled across 20 states, naming transparency failure as the leading cause. Gallup found in May that roughly seven in ten Americans oppose a data center in their area.

The Grade

Mechanism: externally confirmed. The August 10 reporting confirms the qualitative mechanism inside the forecast window. Community sentiment and permitting readiness now function as credit variables in institutional underwriting, as stated nine months in advance. Interpretive confidence: 85–92%.

Quantitative predictions: open. Nothing in the reporting settles the basis-point effects, the scoring thresholds, or the causal magnitude. Project-level pricing differentials are not publicly observable, the share of deal covenants incorporating a consistency measure stays undisclosed, and whatever instrument lenders use is proprietary. Interpretive confidence that the magnitude claims settle within the stated bands: 40–55% — insufficient to claim.

Reporting the numbers as confirmed would overclaim. The mechanism is what the record confirms, and the mechanism is what the rest of this publication builds on.

III. Authorization Has Two Prices

Every data center pays twice for permission. One price is visible in the permit file. The other is visible only in the capital structure, and the second price can exceed the first.

Each price is denominated in the currency of the party charging it, which is why the two rarely meet on the same page.

Read across any row and the translation problem appears. A state imposing a fifteen-year minimum-payment commitment is setting a duration; a lender reading the same term is setting a coverage ratio. Neither party writes down the other’s number.

The proposition joining the two is a translation rather than an accounting identity:

The authorization price becomes a credit spread when authorization uncertainty changes the probability distribution of the cash flows supporting capital.

Five components carry the translation. MindCast groups them as Authorization Credit Exposure (ACX):

ACX = Expected Delay Carry + Expected Cancellation Loss + Refinancing Exposure + Contractual Exposure + Repricing Cost

Expected Delay Carry — interest, leases, labor, equipment commitments, and site costs incurred while productive operation stays blocked.

Expected Cancellation Loss — stranded site development, non-redeployable equipment, terminated contracts, impairment. Blackstone-owned QTS terminated its Prince William Digital Gateway project in Virginia after sustained local opposition, and did not approach lenders for bank financing on the project at all. A privately capitalized developer will end a multibillion-dollar program rather than pay an authorization price that has climbed past the return.

Refinancing Exposure — the risk that delay pushes a project across a maturity, bridge, warehouse, or construction-financing boundary.

Contractual Exposure — penalties and lost economics tied to tenant delivery, power obligations, and dated customer promises.

Repricing Cost — the additional return lenders demand once perceived authorization risk changes.

Calibration must precede any use of ACX as a scoring instrument. The code carries an X rather than an E to avoid collision with an existing equilibrium concept in the MindCast registry — a discipline the firm applies to every new metric before publication. One implication is available immediately:

The same statute imposes different credit prices on different capital structures.

Regulatory economics cannot stop at the legal term. Credit analysis cannot begin at the bond. The transmission mechanism between them is the object worth measuring, and the following sections measure three components of it.

IV. Liquidity Latency Is a Jurisdictional Variable

MindCast defined liquidity latency in November 2025 as the interval between funding commitment and energization, and found that projects energizing within six months command a 10 to 15 percent return premium over projects taking twelve to eighteen. The original formulation treated latency as a property of the developer. Latency is substantially a property of the jurisdiction.

The conversion chain runs in one direction and breaks at any link: site → authorization → interconnection → firm power → construction → energization → utilization → revenue.

Capital enters before the chain completes. Every remaining link therefore carries the possibility that latency moves after the financing assumptions have locked.

A capital committee underwriting a 2028 revenue start does not experience a 2029 start as regulatory friction. The committee experiences another year of carry, another year before depreciation-producing assets generate revenue, a probable refinancing, tenant and lease mismatch, hardware obsolescence risk, and a revised judgment about management’s execution credibility.

Physical constraints set the floor. McKinsey reported in June 2026 that lead times for generators, chillers, transformers, and switchgear have more than doubled since 2019, with some exceeding three years, and that average grid connection waits now exceed four years and reach a decade in some markets. MindCast's own energy work identified transformer manufacturing as the narrowest chokepoint in the entire stack.

Authorization variance sits on top of the floor, and jurisdictions control it directly. Two states with identical grid conditions can produce twelve months of difference through permitting design alone.

Measuring that variance in months is settled work — every state already carries a score. Measuring it in basis points is what this publication attempts, and the three sections that follow identify the exposures that make the measurement urgent.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

V. Curtailment Becomes a Credit Exposure in June 2027

Grid access changed character in 2026. Connection used to be a construction milestone that a project cleared once. Connection is becoming a continuing condition of service that a project can lose, and the financing materials reviewed for this publication do not price the difference.

The PJM board directed two filings on July 27, 2026, submitted to FERC on July 31 under Section 205 of the Federal Power Act. One establishes a Reliability Backstop Procurement addressing capacity shortfalls in the 2027/2028 and 2028/2029 auctions, which PJM attributes primarily to the addition of large loads and data centers. The other places new large loads that neither bring new generation nor otherwise secure supply into curtailment during capacity shortages, ahead of measures affecting other customers.

PJM defines a large load as an end-use customer drawing 50 MW or more at a single interconnection point, or across multiple points within a one-mile radius, and will maintain a registry of them.

Texas began auditing generation plans and ERCOT reliance days later. FERC’s June 18 show cause orders direct all six regional operators to address cost shifting, transmission cost transparency, co-location, and behind-the-meter generation.

Supply-or-curtail changes what a lease is worth.

A lease priced on assumed availability, financed by construction debt serviced from that lease, and rated on contracted revenue now carries a regulatory right of interruption held by a party outside the contract.

The chain is short and each link is a credit concept. Curtailment risk is uptime risk. Uptime risk is revenue risk. Revenue risk is debt service risk.

Public financing materials reviewed for this publication did not identify a dedicated curtailment reserve, curtailment covenant, or curtailment-adjusted debt-service coverage ratio.

Where the exposure lands depends on a lease term nobody has tested. A hyperscaler absorbing curtailment loses compute. A developer absorbing curtailment may breach a service-level agreement to a tenant who then withholds rent that services debt.

Which outcome occurs turns on whether standard triple-net data center leases and hyperscale take-or-pay contracts classify a regional grid capacity shortage as force majeure or as landlord default. Force majeure treatment pushes the revenue risk onto the tenant and largely spares the developer’s debt service. Default treatment routes the loss straight to coverage ratios.

Many leases currently in force predate PJM’s 2026 framework, which leaves the allocation unsettled across a large share of contracted capacity. Counsel reviewing a portfolio should read the force majeure clause before reading the coverage covenant.

Escaping the queue now carries an explicit price rather than an abstract advantage. PJM will reduce the backstop procurement target to reflect documented bilateral contracts and self-supply arrangements delivering net-new capacity for new load, and allocate the residual cost to load-serving entities across a fifteen-year commitment term. PJM opened bilateral matchmaking on June 9, 2026 and expected initial matches in August.

A data center that signs a documented contract for new generation shrinks the procurement and its own zone’s share of the bill. A data center that does not pays through its utility and accepts curtailment exposure on top.

Secured generation therefore stops being an operating choice and becomes a cost-allocation variable that a lender can price at underwriting.

MindCast named secured firm power as the decisive advantage in August 2025, and the CRE simulations found that generation-owning firms outperform power-purchase-dependent competitors whenever utilities face capacity constraints. Supply-or-curtail converts that operating advantage into a financing advantage with a dated deadline attached.

Equipment lead times closed the decision window before the rule takes effect, which means the June 2027 threshold will sort projects that already made their choice.

Interpretive confidence that curtailment enters project credit analysis as an explicit variable within four quarters: 60–70%.

Curtailment is the clearest case in the publication of an authorization-layer rule converting directly into a credit variable. A grid operator, acting under federal review, can now reduce the revenue a financed asset produces — and the contracts governing who absorbs that loss were written before the rule existed.

VI. The Financing Inversion — A Stated Hypothesis

Why the status is a hypothesis. No sponsor states the mechanism and none would, so the read rests on inferred conduct. One transaction carries the behavioral evidence. Project quality, demand strength, and political skill all confound the relationship between leverage and concession. Prediction P14 supplies the test.

Conventional intuition treats money as leverage at a negotiating table, and usually the intuition holds. A counterparty with more cash, better market access, and a stronger balance sheet makes larger commitments and absorbs more adverse outcomes.

Data center construction introduces a second phase where the intuition reverses.

Before commitment, capital expands choice. A firm buys land, orders equipment, reserves generation, negotiates from several sites at once, shifts a project between regions, and credibly threatens to walk.

After commitment, capital contracts choice. Drawn debt accrues interest. Custom facilities resist repurposing. Equipment and power orders become non-cancelable. Tenant dates approach. Local construction begins. Financing maturities keep moving even when permitting stops.

Political delay acquires a price the moment the second phase begins.

The variable governing the reversal is Delay Tolerance, and it rises with liquidity, operating cash flow coverage, undrawn capital, and geographic optionality, while falling with drawn project debt, fixed carry, contractual deadlines, refinancing pressure, and irreversible commitments.

The same capital commitment that increases a firm’s capability to complete a data center decreases its capability to resist delay once the commitment becomes irreversible.

Why Capital Sequencing Matters More Than Previously Stated

MindCast’s CRE simulations found that firms closing financing only after interconnection agreements execute achieve better execution and tighter spreads, and attributed the gain to avoiding idle capital.

The gain is larger than the original attribution captured. Sequencing capital behind authorization preserves the ability to walk away, and the ability to walk away preserves the reservation price in every subsequent negotiation with a state, a utility, or a county.

Timing beats speed for a financial reason and a bargaining reason. Only the first has been named.

Three Postures, Three Clocks

Three architectures dominate the market, and the same authorization shock costs each of them something different.

The bottom two rows carry the finding. State leverage rises as delay tolerance falls, and delay tolerance falls as capital becomes irreversible — which is why the largest sponsor at a hearing is not always the strongest one.

Oracle runs the most exposed position visible. Remaining performance obligations reached $638 billion at the close of fiscal 2026, up 363 percent year over year, against negative free cash flow of $23.7 billion and capital expenditure of $55.7 billion. The company raised $43 billion in debt and $5 billion in equity during the year, absorbed an S&P downgrade to BBB−, and concentrates roughly $300 billion of backlog in a single counterparty. Oracle also disclosed $75 billion of prepaid and customer-supplied hardware — counterparty commitment converting directly into financing capacity.

Meta distributes financing outward while retaining contractual exposure. The Hyperion vehicle raised approximately $29.5 billion in the largest non-merger high-grade bond sale on record, with PIMCO taking roughly $18 billion and Blue Owl Capital participating as joint venture partner on the Louisiana campus. JPMorgan and Morgan Stanley separately managed a $12.3 billion bond sale for BlackRock as Meta’s partner on an El Paso project.

Diversified hyperscalers funding from internal cash occupy the strongest position and face the least pressure to concede.

One transaction shows the mechanism operating early. In December 2025, Blue Owl Capital did not proceed with a contemplated $10 billion financing for the Oracle-linked Saline Township campus in Michigan after the parties failed to reach terms. Reporting by the Financial Times and Reuters attributed the breakdown to tougher leasing and debt terms demanded by lenders, lease economics less favorable than Blue Owl’s earlier Oracle deals, mounting scrutiny of Oracle’s leverage — and concern that local politics in Michigan would delay construction.

Oracle disputed the characterization, stating that its development partner Related Digital had selected a different equity partner from a competitive field and that negotiations were on schedule.

The dispute over the deal’s status does not affect the analytical point. A private credit firm weighed local political delay risk in a funding decision in December 2025, eight months before Reuters reported the practice as general across the lending market.

The Reservation-Price Consequence

A rational firm concedes until the marginal cost of another concession exceeds the expected cost of the delay that concession avoids.

Suppose a state asks for $150 million of additional system investment. The correct corporate question is not whether $150 million is expensive. The question is whether refusing produces an expected delay cost above $150 million.

An identical legal term therefore destroys value at one project and creates value at another. Which outcome occurs depends on the capital structure, not the statute — and the next section follows the consequence to the state side of the table.

VII. The State-Extraction Hypothesis: Why Low Delay Tolerance May Raise the Price a Firm Accepts

A state looking at a large proposed campus sees enormous corporate capability and assumes a difficult counterparty. The assumption may be backwards.

Capability and negotiating flexibility move independently. A $20 billion project carries more economic power than a $2 billion project and simultaneously more exposure to a six-month delay.

States do not need a private financing model to find the difference, because reservation prices leak through conduct.

A developer under timing pressure offers additional infrastructure, accepts longer contractual terms, expands collateral, strengthens community benefits, funds dedicated generation, agrees to enhanced reporting, accepts larger exit protections, or trades economics for a dated path to energization.

A less constrained counterparty declines, resizes, relocates, or waits for the next session.

The hypothesis is testable as written:

Holding project quality and expected demand constant, lower delay tolerance correlates with a higher authorization price accepted by the applicant.

Interpretive confidence in the direction: 75–85%, held below release. The direction is argued, not established.

A correction the simulation forced. Arbitration across the twin network found that structural constraint, not sponsor conduct, is the binding cause of authorization outcomes over the forecast horizon. Physical and capital constraints — transformer lead times beyond two years, interconnection waits beyond four, the June 2027 supply threshold, and capital already drawn — govern outcomes more than sponsor posture does through 2027. A sponsor that decides today to change its conduct still cannot change its lead time, its queue position, or its committed capital.

The tier-one levers in Section XV remain correct and nearly free. Their expected effect over the next eighteen months is smaller than a conduct-dominant reading would suggest.

Financing Posture: The Missing Bargaining Column

MindCast’s bargaining architecture currently scores firms on resources and conduct. Seven inputs determine how much regulatory delay a project can absorb, and none appear in the current framework.

Funding source — operating cash flow, corporate debt, project debt, SPV or JV financing, private credit, customer prepayment, or mixed.

Cash-flow coverage — whether existing operations fund the investment comfortably or the project depends on future earnings.

Draw status — how much project capital remains optional versus already funded or committed.

Duration match — whether financing maturities exceed realistic authorization and construction timelines.

Contract coverage — how much future revenue rests on enforceable customer obligations.

Geographic optionality — whether capital and demand can migrate to an authorized site elsewhere.

Irreversibility — how much committed capital loses value if the project delays, resizes, or moves.

Financing Posture does not replace the existing Capability Bundle. Financing Posture changes what the Capability Bundle means, because a financially powerful firm can score high on capability and low on delay tolerance at the same time.

For a state, the practical instruction is simple: the terms available depend on which sponsor is asking, and the answer sits in the sponsor’s financing structure rather than its public commitments.

VIII. The Limit on State Extraction

Extraction works, and extraction has a ceiling. States that miss the ceiling damage their own future market, and the damage arrives with a lag long enough to escape political accounting.

Rising authorization costs eventually push a project across one of three boundaries.

The return boundary — expected project return falls below the investor’s hurdle rate.

The financing boundary — lenders no longer fund the structure on acceptable terms.

The optionality boundary — another jurisdiction becomes more valuable despite relocation cost.

A state therefore maximizes value below a project-killing threshold. The optimal authorization price is not zero. The optimal price is also not the maximum concession available from the most desperate bidder in the queue.

Repeated extraction above the frontier raises the jurisdiction’s own future cost of capital. Investors learn. Developers reroute pipelines. Tenants demand alternatives. Utilities lose credible load.

MindCast’s Regional Integration analysis documented the dynamic before the current legislative wave. Northern Virginia’s score fell as trust eroded and project timelines stretched from roughly eighteen months toward twenty to thirty-six, while trust-rich regions absorbed the displaced demand. Arizona showed how one constraint failure — water — undermines otherwise strong power infrastructure.

Jurisdictional reputation behaves like a priced asset that depreciates. A state can win a bargain and lose a market in the same session.

IX. The Opacity Loop: SPV Structures Manufacture Permitting Risk

The most damaging structural pattern in the 2026 record is one the industry created for good reasons and cannot escape alone.

Special-purpose vehicles, joint ventures, and layered ownership protect corporate ratings, isolate project risk, and preserve negotiating leverage during site acquisition. Every purpose is legitimate. Every purpose is achieved.

The same structures produce ownership opacity. Carbon Direct examined 46 delayed or cancelled projects worth $170 billion across 20 states and found transparency failure the most common cited cause — specifically nondisclosure agreements signed by elected officials, complex ownership chains, and undisclosed power and water requirements.

Lenders now price the opposition that results, and two disclosures from the August 2026 reporting show the pricing already operating.

CyrusOne carries a $9.7 billion warehouse and revolving credit package arranged by Morgan Stanley, TD Securities, KKR Capital Markets, and Wells Fargo. Part of the facility funds new construction only if all permits and leases are in place. The company faces resident opposition to a $500 million project in Sangamon County, Illinois.

Compliance runs continuously rather than at close. Builders must demonstrate before each drawdown that a project still satisfies the financial covenants and monitoring requirements agreed with lenders. Permitting failure therefore stops capital mid-construction, not only at origination.

The loop closes:

Financing architecture chosen to reduce credit risk manufactures the permitting risk that raises it.

MindCast measured the effect in November 2025 without identifying its source. Phoenix water backlash extended permit timelines nine months; proactive transparency in the Carolinas halved them. Transparency already registered as capital efficiency.

What earlier work treated as a communications variable is substantially a structural one. Ownership arrangements that trigger opposition get chosen in the financing room, not the community affairs office.

Why Advice Will Not Fix It

Naming the loop as an equilibrium rather than a mistake determines the remedy, and the distinction is not academic.

No single sponsor discloses its ownership chain first. Unilateral transparency surrenders negotiating position while competitors stay opaque, and delivers a benefit — lower spread — that arrives only once disclosure becomes general. Non-disclosure holds as a stable best response for every player.

Equilibria of that shape break only under force applied to all sponsors at once. Three candidates exist: a lender covenant applied uniformly, a state statute compelling identification of the ultimate parent, or a rating agency criterion naming ownership disclosure as a rating factor.

Compelled disclosure costs developers something real, and less than they assume. Naming the ultimate parent on an application invites land-cost inflation, since sellers price against a known deep-pocketed buyer. Set that cost against the alternative. Permitting delay accrues continuously against drawn capital, while a land premium is paid once at acquisition. A universal rule also equalizes the penalty, since every competitor faces the same seller.

Texas ran the experiment in an adjacent field. The state asked data centers to report power and water use. Fewer than ten percent complied. Governor Abbott then halted approvals inside the ERCOT interconnection process and made the audit a condition of grid access.

Voluntary disclosure failed. Attaching grid access as a consequence changed the obligation from a request into a condition of project progression. An equilibrium reading predicts exactly that sequence; an error reading does not.

Texas carries a second lesson for capital. Approvals stopped with no stated end date, under public federal pressure, in the jurisdiction that had served as the permissive benchmark. A sponsor cannot price an indefinite pause, and unpriceable variance in a welcoming state deserves more attention than demanding terms in a hostile one.

Interpretive confidence that the loop operates as described: 70–80%. Confidence that it survives testing against opposition rates sorted by ownership structure: 60–70%.

The practical reading for an investor is short. A sponsor cannot fix ownership opacity alone without losing ground to peers who do not, so the fix arrives from a lender, a rating agency, or a statute — and whichever arrives first resets the cost of capital for every sponsor at once.

X. The SEC Staff Reclassification of Data Center Securitizations

A staff action in late July 2026 removed the mandatory alignment backstop from a fast-growing data center financing channel. The timing, rather than the action itself, is what deserves attention.

On July 29, 2026, the SEC's Division of Corporation Finance issued a staff interpretive letter responding to a request from Latham & Watkins, concluding that securities issued in the described data center securitizations are not asset-backed securities under Section 3(a)(79) of the Securities Exchange Act. The staff reasoned that data center facilities are tangible operating assets enduring beyond the tenor of the securities, rather than self-liquidating financial assets.

Characterize the action precisely. A staff interpretive letter is not a Commission rule or statement and carries no independent legal force. The letter binds nothing and reflects the staff’s view on the facts presented.

The practical consequence still reaches Regulation RR risk retention, Rule 192, Rule 15Ga-1, and Rule 15Ga-2, which generally require a five percent retained interest for covered asset-backed securities.

Mandatory federal risk retention no longer supplies the alignment backstop for structures falling within the staff interpretation.

Retention exists to align sponsor and investor interests, and the backstop came off in the same month lenders began pricing permitting status and community opposition as credit variables.

Scale makes the timing consequential. Data center asset-backed and mortgage-backed issuance passed $25 billion in 2025, up from $2.4 billion in 2020, and 2026 is pacing toward another record. MindCast forecast in November 2025 that green and infrastructure-backed securities would reach $150 to $200 billion in annual issuance through 2028. Regulatory friction has now been removed from the vehicle that forecast identified.

Two considerations bound the concern. Sponsors have typically retained around 30 percent of these structures against a 5 percent requirement, so the near-term practical effect may prove modest — market demand, not regulation, has been the binding constraint. The guidance also removes genuine compliance friction that participants argued never fit the asset class, and lower friction lowers the cost of infrastructure the country has decided it wants.

A second collateral question arrived the same week. On August 10, 2026, Nvidia announced memorandums of understanding with six major financial institutions to mobilize more than $500 billion in third-party capital for AI infrastructure, structured as long-term debt and asset-backed financing for customers buying GPUs and building data centers. Jensen Huang characterized the platform as addressing circular-financing concerns directly, stating that the capital is not Nvidia revenue and that demand is real. The initiative targets precisely the sponsors this publication identifies as least able to wait — buyers without the credit rating or cash to purchase silicon outright. Two collateral assumptions now sit underneath the same financing structures. The first is that a data center's authorization will hold, which this publication argues is underpriced. The second is that GPUs retain value over a debt tenor, which analysts have questioned on depreciation grounds and which Nvidia counters by pointing to software updates preserving useful life. Nvidia's five-year credit default swap spread roughly doubled between late May and August 10. A project can service its debt only if both assumptions hold, and neither is settled.

The forward risk is the combination, not the letter. A financing channel growing rapidly. Sponsor alignment now voluntary for qualifying structures. Ownership structures that generate the opposition lenders have just started pricing. And a curtailment regime arriving in June 2027 that no reviewed offering document addresses.

Each element is defensible alone. The four together describe a channel where authorization risk migrates toward the holders least equipped to observe it.

Status of the consequence. The regulatory fact is settled and dated. The consequence is not. Thirteen days is insufficient time for observable sponsor behavior, and the voluntary retention noted above remains a live alternative explanation. The combination is monitored, not forecast.

Interpretive confidence that the combination raises system-level exposure: 65–75%, held below release for insufficient causal integrity. Prediction P15 supplies the observable test.

A fixed-income holder buying into this channel should ask one question the offering document may not answer: how much of the deal is the sponsor keeping, and why.

XI. Two Instruments: Financeable Authorization Coverage and Capital Irreversibility

Credit markets need a way to separate capital committed behind mature authorization from capital committed behind hope. Two ratios do the work, and the pair matters more than either alone.

Financeable Authorization Coverage (FAC) divides financially committed megawatts that have cleared every material authorization gate by total financially committed megawatts. The numerator is a subset of the denominator by construction, so the ratio necessarily runs from 0 to 1. The name avoids collision with an existing conversion metric already registered under the ACR code.

A project financing 1,000 MW with 800 MW through all material gates carries a FAC of 0.80.

Thresholds. Low FAC is below 0.50. High CIR is above 0.60. Both govern the settlement of Prediction P13. A company announcing 10 GW while committing capital only after approvals retains strong coverage. A company committing billions against early-stage sites carries weak coverage despite an impressive pipeline.

Materiality governs the FAC numerator, and the rule is binary rather than proportional. Any unresolved critical-path gate — power service, interconnection, zoning, environmental authorization, water allocation, or a state construction prohibition — drops the associated capacity out of the numerator entirely. A project holding an executed interconnection agreement but missing a county water allocation scores that capacity at zero, not at a partial credit.

Non-critical items score nothing against the ratio. A pending landscaping permit does not reduce coverage, because a landscaping permit cannot stop energization.

The Model Authorization Code supplies the hierarchy that separates the two categories; the financial model supplies the denominator.

Capital Irreversibility Ratio (CIR) divides non-cancelable or materially stranded capital by total project capital.

Numerator rule. The numerator counts capital already disbursed, plus contractual commitments carrying a termination penalty above 25 percent of committed value, plus site-specific works with no redeployment path. Equipment orders transferable to another site of the same sponsor do not count.

Together the two ratios sort projects into four states.

The fourth quadrant should concern capital providers and governments equally. Quadrant four identifies the project most likely to convert an authorization shock into a credit event, and simultaneously the project from which a state can extract the most.

Lenders already enforce a crude version of the ratio. The CyrusOne facility releases construction capital only against completed permits and leases, which is Financeable Authorization Coverage applied as a covenant rather than a score. Formalizing the measurement lets a creditor compare projects rather than police one.

Both ratios measure a project. MindCast’s earlier work measured firms, asking whether an institution executes what it says. The two questions are different and both matter: a disciplined sponsor can still hold a project that has not earned the right to be built.

Neither instrument is validated. Prediction P13 tests whether the pair actually sorts risk, and a failure there retires both.

XII. Delay Tolerance Enters the Bargaining Equation

MindCast’s existing framework scores a firm’s bargaining position at a state table on two things: the capability it brings — capital, clean power, demand credibility, execution record — and a discount for past conduct, since a sponsor that broke a prior commitment gets worse terms. Call the first the capability bundle and the second the conduct discount.

The finance layer adds a third term the framework did not carry. Time.

Delay tolerance resolves into one classification:

High — the project can defer, relocate, resize, or absorb a prolonged authorization period without material financing impairment.

Medium — delay raises cost and may alter returns without threatening financing viability.

Low — material delay creates refinancing, covenant, contractual, liquidity, or impairment risk.

Delay tolerance enters as a separate multiplier rather than as part of capability:

Effective Bargaining Position ∝ Capability Bundle × Delay Tolerance × (1 − Conduct Discount)

Geographic optionality — whether the sponsor can move the project elsewhere — sits inside delay tolerance rather than beside it, which keeps the expression to three terms.

Calibration remains pending, and the correction stands without it. A sponsor with the strongest capability in the market can still bargain from the weakest clock, which is why capability alone predicts terms poorly.

XIII. Creditors Hold the Exposure and Have No Seat

Analysts describe data center regulation as a contest among developers, hyperscalers, utilities, regulators, lawmakers, local governments, and communities. The description is no longer complete.

A growing share of the economic exposure now sits with bondholders, infrastructure funds, pension funds, insurers, private credit lenders, and securitization investors. Banks reached concentration limits and passed the paper onward to exactly those holders.

Banks already fold permitting readiness and local political risk into underwriting. The July 29 SEC staff letter opens another pathway for infrastructure cash flows to reach fixed-income markets, and opens it with sponsor retention now voluntary for qualifying structures.

Creditors do not sit at the state authorization bargaining table. Their influence arrives indirectly, through credit agreements, draw conditions, required returns, lease structures, financing maturities, and willingness to fund the next project.

Foreign and sovereign capital carries the exposure in its sharpest form. A pension fund holding project debt is anonymous to a county hearing. A sovereign wealth fund or foreign strategic investor is not, because ownership identity becomes an argument at the hearing rather than a line in a filing.

The consequence inverts the remedy. Disclosing the ultimate parent lowers a domestic sponsor’s capital price by removing the leading cited cause of cancellation. Disclosing a foreign or sovereign parent can raise the authorization price instead, by handing organized opposition its most effective argument.

Sovereign and foreign allocators therefore face a disclosure equilibrium that binds harder than it binds anyone else, alongside national-security review that domestic capital does not encounter. Both effects compress the jurisdictions where that capital can deploy at an acceptable authorization price, and neither appears in conventional country-risk analysis.

Interpretive confidence in the asymmetry: 65–75%.

Three parties ask three different questions about the same regulatory event.

A government affairs team asks what the state will require. A developer asks what to concede. A creditor asks:

How does the state’s next move change the probability distribution of the cash flows supporting my capital?

Political risk analysis rarely reaches the data center tariff clause. Credit analysis rarely models the state legislative bargaining that produces it. Bankers leaning toward welcoming states are running an unmeasured version of a fifty-state comparison that already exists in published form — which is the commercial gap this publication identifies.

Does Financing Volume Rise or Fall From Here?

Every allocator reading the analysis above asks the same question, and the framing hides the useful answer. Volume rises. Velocity falls. Dispersion widens. Three separate movements, and only the third is investable.

Volume rises, and not because capital expenditure rises. The funding mix is shifting from operating cash flow toward external capital. Debt already funds roughly a third of hyperscaler capital expenditure and is heading toward 35 percent against $1.14 trillion of spending in 2027. A flat capex year still produces more issuance under that shift. Securitization compounds the effect: data center asset-backed and mortgage-backed volume climbed from $2.4 billion in 2020 to past $25 billion in 2025, and the July 29 staff letter lowered the compliance cost of that channel in the same month authorization risk entered credit analysis.

Velocity falls, which is the part most forecasts miss. Permit-to-draw conditions leave committed capital undrawn. Four categories that used to move together have separated — announced, committed, drawn, and deployed — and a project can sit in the second for years. Estimates already place $150 to $200 billion of 2026 spending slipping into 2027 and 2028.

Dispersion widens, and dispersion is where the money is. Authorization risk does not shrink the total. Authorization risk reroutes it. Projects carrying a complete authorization stack finance faster and cheaper. Projects without one finance slower, wider, or not at all. QTS terminated Prince William without approaching lenders. Blue Owl declined the Michigan financing while continuing to fund other campuses for the same sponsor. Neither event reduced sector issuance; both moved it.

What would actually reverse the volume. Permitting fights cannot shrink the aggregate on their own — they change who gets financed and at what price. Only a demand revision does that. Three indicators settle the question in order: contracted backlog quality at the concentrated sponsors, capital expenditure guidance at the four largest buyers, and whether pre-lease ratios hold near current levels.

Interpretive confidence in the volume-up, velocity-down, dispersion-widening read: 70–80%. Confidence that a demand revision rather than an authorization event is the binding reversal risk: 65–75%.

An allocator worried about the sector shrinking is watching the wrong variable. The sector bifurcates well before it contracts.

XIV. Capital Markets Settle Predictions Every Quarter

The finance layer delivers a methodological benefit that has nothing to do with the subject matter.

Legislative predictions settle slowly. A session, an election, a commission case, or a multiyear build cycle can take eighteen months to resolve.

Capital markets settle every quarter. Capital expenditure guidance, free cash flow, debt issuance, spreads, credit-default swaps, rating actions, remaining performance obligations, backlog, lease commitments, financing closings and cancellations, construction slips, impairments, depreciation changes, and authorization disclosures all arrive on a fixed clock against named sources.

Pairing the two produces separate contracts from a single event. A regulatory prediction names a jurisdiction changing the authorization price. A capital prediction names an exposed firm changing financing, disclosure, project timing, or contractual structure within a specified number of quarters.

The two settle independently. A regulatory hit cannot rescue a financial miss, and a financial hit cannot retroactively validate a regulatory miss.

Section II applies the discipline to MindCast’s own record, where the mechanism settled and the magnitudes did not. Separation of that kind strengthens a falsification architecture rather than diluting it.

XV. Recommended Actions by Stakeholder

Each party can act on the analysis above without waiting for the others, and the highest-value actions differ sharply by role.

Hyperscalers

Treat authorization maturity as a capital allocation variable before committing irreversible spending.

Build geographic substitutability, which functions as regulatory insurance and protects the reservation price at the same time.

Disclose the development pipeline by authorization maturity rather than announcement value — announced, site-controlled, power-secured, materially authorized, under construction, energized. A portfolio of 20 gigawatts of announced projects does not represent 20 gigawatts of equal economic probability.

Disclose capital at risk before the next gate. Total capital expenditure answers the wrong question. Investors need to know how much capital becomes impaired if the next regulatory gate fails.

Developers

Secure material authorization earlier, because entitlement maturity is financial inventory. A permitted, power-secured site holds value when construction markets weaken; an attractive site without a credible authorization path may hold very little.

Maximize pre-lease and counterparty coverage, so lenders underwrite one risk rather than three.

Finance in tranches, so capital becomes non-cancelable only as project probability rises.

Negotiate state terms before financing closes. A developer who closes financing before state obligations stabilize hands government an option on its own balance sheet.

Name the ultimate parent on the initial application. The filing itself is inexpensive, although early parent disclosure can raise land-acquisition costs, and severing the opacity loop at its origin is worth the premium.

Lenders and institutional investors

Underwrite the authorization stack directly rather than substituting a hyperscaler tenant’s credit rating for project authorization analysis.

Measure capital exposed before the next unresolved gate.

Distinguish legal ring-fencing from economic recourse, and ask what voluntary sponsor retention survives the July 29 staff letter.

Stress-test delay rather than only cancellation, and ask what curtailment exposure the structure carries after June 1, 2027.

State lawmakers

Publish predictable authorization floors, and resolve grandfathering and transition rules before capital commits.

Separate cost causation from discretionary extraction.

Require identification of the ultimate parent in siting applications. The requirement costs the state nothing, breaks an equilibrium no sponsor can break alone, and addresses the leading cited cause of project cancellation.

Recognize that indefinite pauses cost sponsors more than demanding terms, and that uncertainty raises the financing price without changing the nominal legal burden.

Federal lawmakers and regulators

Recognize that fragmented state authorization now transmits into national infrastructure financing.

Federal acceleration that leaves downstream state and local variance untouched moves the bottleneck without reducing its capital cost.

Design federal performance standards and transmission policy around the financeability of the whole authorization stack rather than federal permit speed alone.

Across all five audiences the instruction is the same one stated differently. Find out what permission actually costs, in months and in basis points, before the capital becomes impossible to move.

XVI. How MindCast Generated Foresight Predictions

MindCast produced the register below by running the analysis as a MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation (MP CDT FS) rather than by expert judgment. A reader deciding how much weight to place on a number deserves the method first.

System provenance. The register includes outputs generated through an independent Claude implementation of the MindCast MP CDT FS runtime.

The simulation builds behavioral models of the actors, not summaries of them. Nine Cognitive Digital Twins were instantiated for this run: a state government twin, a grid operator twin, a local government twin, a community coalition twin, three developer twins representing balance-sheet-first, contract-first, and independent capital architectures, a bank lender twin, and a composite creditor twin covering private credit, rating agencies, and securitization investors. A system twin carries the governance equilibrium state. Each twin encodes objectives, constraints, beliefs, incentives, available options, installed decision grammar, adaptation velocity, pressure thresholds, response policy, observable signals, failure modes, and its own falsifier.

Four explanations competed, and the run chose among them. Structural constraint, strategic use of delay, incentive response, and institutional coordination each offer a complete account of why sponsors concede to states. The simulation forced them to compete rather than blending them into a synthetic score, and where two remained viable it branched instead of averaging. Section VII reports the result.

Cognitive Digital Twins play against each other, never against themselves. The unit of foresight is adversarial cross-simulation. A developer twin bargains against a state twin and separately against a lender twin. A large-load twin contests a grid operator twin. A developer twin faces a community coalition twin. Every recorded interaction pairs distinct actors in contest, and a run without executed cross-twin interaction is analysis rather than simulation.

Causal claims pass an integrity gate before they can carry a prediction. Each claim scores on whether visible action honestly expresses the underlying decision system, whether stated commitments translate into observable behavior, and whether the signal holds across time and across actors — penalized for confounding. A claim scoring below the release threshold cannot support a released forecast regardless of how compelling the argument reads.

Termination requires two independent conditions. Behavior must converge — no actor improves by unilaterally changing strategy — and further search must stop adding enough information to justify its cost. Convergence alone does not close a run.

Predictions lock before the simulation ends, not after. Inputs lock first, routing next, predictions last. Every entry below settles against a named source.

What the method cannot do. The simulation tests whether a specified architecture executed faithfully and whether claims survive the integrity gate. Predictive accuracy settles only against the public record, separately and later. A well-executed run is not a correct one, and MindCast grades the two on separate ledgers.

XVII. Foresight Prediction Register

Every entry carries a settlement source, an explicit falsifier, a dated window, and whether the claim governing it cleared the causal-integrity gate. Event probabilities and interpretive confidence never pool.

Version 1.0: thirteen event predictions grouped by the signal they generate, two hypotheses under test, three structural reads.

Reading the causal-integrity line. Cleared entries carry probabilities. Entries below the threshold carry none, and appear as hypotheses with tests attached.

The definitions below govern every settlement. Top-ten project means a 2026-announced US project ranking in the top ten by announced IT capacity in megawatts. Additional state means a state without an approved mandatory large-load tariff as of August 11, 2026. Historical retention practice means the median disclosed sponsor retention across data center securitizations closing between January 1, 2024 and July 28, 2026. Delay tolerance classifies objectively: low where any of trailing-twelve-month free cash flow is negative, construction debt is drawn against the parcel, or a contractual delivery date falls within twenty-four months; high where free cash flow is positive, project capital remains undrawn, and two or more alternative announced sites sit in different power markets; medium otherwise.

Credit Documentation — What Becomes Standard Covenant Architecture

P1 — Permit-to-draw conditions spread. Three or more data center financings originated after August 11, 2026 disclose construction-draw conditions tied to permits, leases, or authorization milestones. Window: by June 30, 2027. Falsifier:fewer than three disclosed. Settlement: offering documents, SEC filings, reporting in Reuters, Bloomberg, or IFR. Note:the CyrusOne package predates this publication and does not count. Causal integrity: cleared, strongest tier. Event probability: 65–75%.

P2 — Rating methodology names authorization. At least one of S&P, Moody’s, or Fitch publishes criteria or sector commentary naming permitting status or community opposition as an explicit rating factor for data center project finance.Window: by December 31, 2027. Falsifier: no such publication. Settlement: the agency’s published criteria and commentary archive. Causal integrity: cleared, strongest tier. Event probability: 60–70%.

P3 — Curtailment named as risk. Curtailment exposure appears as a named risk factor in at least one data center financing document or rating report. Window: by December 31, 2027. Falsifier: no such appearance. Settlement: offering documents, SEC filings, rating agency reports. Causal integrity: cleared. Event probability: 60–70%.

P4 — Curtailment structured, not just disclosed. A data center financing discloses a curtailment reserve, curtailment covenant, or curtailment-adjusted coverage ratio. Window: by December 31, 2027. Falsifier: risk-factor language only, with no structural provision. Settlement: offering documents, rating agency reports. Causal integrity: cleared. Event probability: 40–50%.

Capital Markets — Where the Money Moves

P5 — Securitization volume rises. Data center asset-backed and mortgage-backed issuance in calendar 2026 exceeds the 2025 total. Window: reported by February 28, 2027. Falsifier: 2026 issuance at or below 2025. Settlement: Asset Securitization Report, Bloomberg issuance data. Causal integrity: cleared. Event probability: 75–85%.

P6 — Concessions become investor evidence. At least two sponsors cite a ratepayer-protection agreement, generation commitment, transmission contribution, community benefits agreement, or authorization certainty as evidence of reduced execution risk in an earnings call, investor presentation, financing document, or company release. Window: by December 31, 2027. Falsifier: fewer than two such citations. Settlement: earnings transcripts, investor presentations, offering documents, company releases. Causal integrity: cleared. Event probability: 65–75%.

Authorization and State Response — What the Rule-Setters Do Next

P7 — FERC approves the PJM backstop. FERC approves PJM’s Reliability Backstop Procurement in substantially the filed form. Window: by December 31, 2026. Falsifier: rejection, or approval requiring restructuring that removes the bilateral offset mechanism. Settlement: FERC eLibrary, docket ER26-3380. Causal integrity: cleared. Event probability: 70–80%.

P8 — Bilateral capacity becomes credit evidence. At least two large-load customers publicly disclose a bilateral capacity contract or self-supply arrangement citing PJM cost allocation or curtailment avoidance. Window: by June 1, 2027. Falsifier: fewer than two disclosures. Settlement: PJM filings and match disclosures, company disclosures, earnings materials. Causal integrity: cleared. Event probability: 60–70%.

P9 — Ownership disclosure becomes compulsory. At least one state enacts a statutory requirement that data center siting applications identify the ultimate parent or beneficial owner. Window: by adjournment of 2027 regular legislative sessions. Falsifier: no enactment. Settlement: enacted session law. Causal integrity: cleared. Event probability: 55–65%.

P10 — Large-load tariffs spread with long terms. At least one additional state enacts or its commission approves a mandatory large-load tariff carrying a minimum term of ten years or longer. Window: by December 31, 2027. Falsifier: no new enactment or approval meeting the term threshold. Settlement: enacted session law, commission orders. Causal integrity: cleared. Event probability: 65–75%.

P11 — A state prices financing readiness. At least one state, public utility commission, or major local authorization process expressly uses applicant financing readiness, financial assurance, credit quality, or demonstrated capital commitment as a project-selection variable beyond conventional security requirements. Window: by December 31, 2027. Falsifier: no such use. Settlement: statute, tariff, commission order, request for proposals, executive directive, executed development agreement. Causal integrity: cleared. Event probability: 55–65%.

Project Outcomes and Instrument Validation

P12 — A major project dies on authorization. At least one top-ten US data center project announced in 2026 is publicly cancelled, terminated, or indefinitely paused, with local opposition or permitting named among the reasons. Window: by December 31, 2027. Falsifier: no such event, or cancellation attributed solely to demand or supply chain. Settlement:company announcements, Data Center Watch tracking, county records. Causal integrity: cleared. Event probability: 75–85%.

P13 — The instruments sort risk. Among large US projects announced in 2026, those scoring FAC below 0.50 while scoring CIR above 0.60 experience repricing, restructuring, material delay, or cancellation at a higher rate than projects in the other three quadrants. Window: by December 31, 2027. Falsifier: comparable or lower adverse-event rates in the hold-up-exposed quadrant. Settlement: MindCast project dataset scored against company announcements, offering documents, rating actions, and Data Center Watch tracking. Causal integrity: cleared. Event probability: 60–70%.

P13 is the instrument-validation entry. A failure retires Financeable Authorization Coverage and Capital Irreversibility rather than qualifying them.

Hypotheses Under Test

P14 — Delay tolerance predicts concession. Among a matched sample of major projects receiving material new authorization terms, sponsors classified low on delay tolerance accept higher normalized authorization prices than sponsors classified high.

Matching rule. Projects match on announced capacity within one standard deviation, on power market, and on announcement year. Delay tolerance classifies by the definition above. Normalized authorization price sums committed system investment, guaranteed minimum payments, exit protection, and community benefit obligations, divided by contracted megawatts. Observation closes December 31, 2027.

Falsifier: comparable or lower normalized concession by low-tolerance sponsors under the rule above. Settlement:MindCast project dataset against executed development agreements, tariff filings, community benefit agreements, and disclosed financing posture. Causal integrity: below the release threshold — held as hypothesis. The Financing Inversion stands or falls here.

P15 — Retention divergence. Data center securitization issuance rises after the July 29 SEC staff letter while median disclosed sponsor retention on structures closing after that date falls below the historical practice defined above. Window:by December 31, 2027. Falsifier: post-letter median retention at or above the historical median. Settlement: offering documents, issuance data, rating agency reports. Causal integrity: below the release threshold — held as hypothesis.

P16 — GPU residual value enters credit documentation. At least one financing originated under the Nvidia-arranged platforms discloses a residual-value assumption, GPU depreciation schedule, or collateral revaluation trigger.
Window: by December 31, 2027. Falsifier: no such disclosure. Settlement: offering documents, SEC filings, rating agency reports. Causal integrity: cleared.
Event probability: 55–65%.

Structural Reads — Interpretive Confidence, Never Pooled With the Above

S1 — Certainty beats permissiveness. Capital concentrates in jurisdictions publishing bounded, calculable terms over jurisdictions offering low nominal burden with unresolved rules. Observable: pipeline redirection following a jurisdiction’s terms becoming explicit. Interpretive confidence: 70–80%.

S2 — Geometry governs through 2027. Physical and capital constraints — lead times, queue position, drawn capital — govern authorization outcomes more than sponsor conduct over the next eighteen months. Observable: conduct improvements failing to compress timelines in constrained corridors. Interpretive confidence: 60–70%.

S3 — Extraction frontier crossing. At least one jurisdiction extracts past the point where the extraction raises its own future cost of capital. Observable: developer pipeline redirection following a named extraction event. Interpretive confidence: 55–65%.

Investor Transmission Cards

Predictions state what happens. A transmission card states how it reaches a position. Each card traces a trigger to an observable an investor can check on a dated schedule.

The same scoring produces four exposure states, which are positions rather than recommendations.

Favorable — high authorization maturity, low irreversibility, strong geographic optionality.

Watch — high authorization maturity, high irreversibility. Risk retired, flexibility spent.

Speculative — low authorization maturity, low irreversibility. Contained, and not yet earning.

Hold-up exposed — low authorization maturity, high irreversibility. The quadrant where an authorization shock most likely becomes a credit event, and where a state holds the most leverage.

Falsification Contract For the Publication

The confidence constraint fails if authorization uncertainty shows no measurable relationship to financing cost or structure across the 2026–2027 cohort — specifically, if projects carrying documented organized opposition price at spreads statistically indistinguishable from unopposed comparables matched on size, sponsor, and region.

Revisions preserve the original entry and carry its identifier.

XVIII. Corrections and Limits

MindCast publishes corrections at the same level as findings.

The run downgraded two draft claims. MindCast executed the simulation before publication rather than after, and two claims fell below the release threshold. The run also found structural constraint binding over sponsor conduct, which reduces the near-term effect of the developer actions in Section XV — those actions remain correct and nearly free, and will move outcomes less over eighteen months than an intent-based reading predicts.

Correction. An earlier working formulation of this analysis treated the organizational separation between government affairs and treasury as an unrecognized industry failure. The claim overstated the case. Power Brokers and Digital Real Estate documented in November 2025 that high-coherence developers had already built community affairs divisions staffed by former utility regulators and municipal planners, with community liaisons reporting directly to chief financial officers because trust velocity affects capital access. The organizational remedy was identified before this publication. Adoption breadth remains unmeasured.

Three limits bound the analysis.

The capital price is inferred, not observed. Project-level spreads on private credit and 144A issuance are not public at the granularity the argument would ideally use, so the analysis reasons from issuance structure, rating actions, and disclosed terms. The same limitation constrains the November 2025 magnitude forecasts, which is why Section II grades the mechanism and withholds the numbers.

The extraction inversion rests on direction, not magnitude. One transaction supports a directional argument. Confidence bands throughout mark the difference between a mechanism identified and a mechanism measured.

Timing is unresolved. Lenders have begun pricing authorization risk. Whether the pricing hardens into standing covenant language, softens as competition for AI exposure intensifies, or fragments by region stays open, and the register above exists to settle it.

Conclusion — Authorization Becomes a Component of the Cost of Capital

The AI data center buildout began as a technology story. Compute demand collided with the grid and made it an energy story. Power scarcity, ratepayer exposure, water use, and local opposition forced governments to ration permission and made it a regulatory story.

The buildout is now a credit story, and MindCast forecast the transition nine months before the reporting confirmed it.

Debt, leases, private capital, SPVs, project structures, securitizations, and customer commitments increasingly finance the physical layer. Once they do, liquidity latency enters the capital structure and authorization delay begins accruing against drawn money.

A legislative session alters a financing model. A utility tariff changes fixed obligations. A community fight moves a construction draw. A curtailment rule taking effect in June 2027 reaches leases signed today.

Permission therefore carries a second price. The Authorization stream measured what states charge. The Capital stream measured what institutional coherence earns. Joining them answers a question neither could reach alone:

The authorization price becomes a credit spread when regulatory uncertainty changes the timing, probability, or recoverability of the cash flows supporting capital.

The reciprocal proposition may prove more consequential:

Capital structure determines how much authorization price a firm can rationally resist.

Financial capability gets a project to the table. Financial irreversibility keeps it there.

Firms that grasp the distinction can build capital structures preserving bargaining power and investor confidence at once. States that grasp it can write demanding rules without making projects unfinanceable. Creditors that grasp it can tell a billion-dollar facility apart from a billion-dollar claim on permission not yet secured.

The next unit of account in AI infrastructure is not another measure of capital expenditure. The unit is financeable certainty.

Analysts must therefore ask not only what a state will charge, but which counterparty has already committed capital in a way that removes its ability to refuse.

Appendix A — Published Theoretical Foundations

Four published MindCast frameworks supply the theory the simulation runs on. Each is public; the runtime architecture that applies them is not.

Chicago School Accelerated (December 2025) — Supplies the foundation for treating authorization as bargaining rather than administrative process. Coase separates coordination costs from transaction costs. Becker explains ownership opacity as locally rational conduct while opacity carries no price, and predicts reversal once credit committees attach one. Posner supplies the lag between financing cycles and legal correction.

The Dual Nash-Stigler Equilibrium Architecture (January 2026) — Nash identifies where strategic actors stop improving through unilateral deviation, and supplies the equilibrium reading of the opacity loop in Section IX that separates a stable best response from a correctable error. Stigler governs diligence sufficiency — how far a lender searches community and permitting evidence before the marginal information stops justifying its cost — which is a separate question from the underwriting criteria themselves.

MindCast AI Field-Geometry Reasoning (January 2026) — Treats strategy as movement through a constraint field where timing, capital, and switching costs determine which paths survive. Before commitment a developer holds multiple routes: relocate, resize, delay, restructure, substitute power, or walk. As capital becomes irreversible those routes steepen and some close. Debt does not eliminate agency; debt prices agency.

The Dynamic Predictive Game Theory Collection (July 2026) — Models systems where events rewrite the payoff structure rather than change one variable inside a fixed game. A tariff decision, a debt draw, a tenant commitment, a downgrade, and a moratorium each change the game, which is why delay tolerance must update continuously rather than score once.

Appendix B — The Authorization Corpus

Each entry below supplies a proposition this publication extends.

The MindCast AI Data Center Record — Twelve months of dated claims graded against the public record, with three published corrections.

The Authorization Market — Permission as a scarce economic good, with predictable terms outperforming nominal permissiveness because capital can underwrite them. Extended here: what happens once the price enters a capital structure.

The 50-State Authorization Price Baseline — Strictness, certainty, and authorization friction separated across every state. Extended here: certainty and friction become inputs to Authorization Credit Exposure (ACX).

The 50-State Regulatory Atlas — The standing jurisdictional record. Extended here: the dataset bankers currently approximate by intuition.

The Model AI Infrastructure Authorization Code — Scores any data center law on obligation, consequence, and reach. Extended here: supplies the materiality hierarchy inside Financeable Authorization Coverage.

Three Competing Governance Equilibria — Stable arrangements the contest converges toward. Extended here:Financing Posture changes which equilibrium clears.

The Two-Ledger Data Center Bargain — What a project contributes measured against what it consumes. Extended here: a concession that reduces delay variance operates as credit enhancement.

The Federal-State AI Infrastructure Collision — Nine dated checkpoints; cost allocation settled five months early at FERC. Extended here: multi-layer timing becomes a financing-duration problem.

The Power Stack — Bottleneck power migrating toward actors controlling scarce infrastructure. Extended here: capital structure governs how shocks propagate.

The Bottleneck Hierarchy — Secured firm power decides the buildout. Extended here: supply-or-curtail converts that operating advantage into a financing one.

Appendix C — The Investor and Capital Corpus

Capital Is the New Computing (November 2025) — Establishes capital permanence, names liquidity latency, and identifies the speed, scale, and duration mismatches between project finance and AI timelines. Extended here: liquidity latency is jurisdiction-dependent, and the jurisdictional component is measurable.

Power Brokers and Digital Real Estate (November 2025) — Establishes that lenders underwrite energization timing as rigorously as tenant creditworthiness. Extended here: Section II grades the forecast against the August 2026 record.

Building CRE Coherence for AI Infrastructure (November 2025) — Models twelve platforms across six constraint scenarios and documents capital sequencing discipline. Extended here: sequencing preserves the reservation price, not only the return.

The AI Infrastructure Energy Opportunity Landscape (March 2026) — Introduces Capital Vision, separating constraint-removal capital from choke-point capital. Extended here: the same test applies to authorization strategy.

Executive Summary of the MindCast AI Investment Series (September 2025) — Five studies establishing trust and foresight integrity as quantifiable assets. Extended here: firm-level coherence complements the project-level ratios in Section XI.

Critical references: The Model AI Infrastructure Authorization Code · The Data Center Authorization Market: A 50-State Regulatory Atlas

Executive Summary

On November 16, 2025, MindCast AI published a forecast describing what would follow the Department of Energy’s directive treating hyperscale computing as a matter of interstate grid reliability. Federal regulators would move fast. State regulators would object that federal authority was intruding on their jurisdiction. Companies would pursue their own generation. Transmission cost responsibility would become the central economic fight.

Thirty-nine days later, the outside world produced the first confirmation. Over the following eight months it produced four more. The sequence below runs in the order the evidence arrived.

December 26, 2025 — thirty-nine days. The Wall Street Journal reported state regulators invoking the 1935 Federal Power Act, former federal energy officials warning of litigation, and Florida moving to bar utilities from charging households for hyperscale development. Three of the four predicted dynamics appeared in a single article.

May 13, 2026 — community consent. Gallup found roughly seven in ten Americans oppose an AI data center in their area. MindCast named community trust as the decisive site-selection filter in September 2025, before the first state moratorium and before any national polling existed on the question. Opposition groups have since more than doubled to 833 across 49 states, and both major-party nominees for governor of Ohio have pledged a moratorium.

June 18, 2026 — cost allocation. The Federal Energy Regulatory Commission ordered all six FERC-jurisdictional regional grid operators to justify or reform how large loads connect to the grid, naming cost shifting and transmission cost transparency explicitly. The January 2, 2026 checkpoint required cost allocation to surface in PJM or MISO by late 2026. Confirmation arrived five months early.

June 30, 2026 — physical constraints. The McKinsey Global Institute reported equipment lead times more than doubled since 2019 and grid connection waits exceeding four years. MindCast identified transformer scarcity and transmission congestion as documented constraints in March 2026, while separating them from unsupported claims of imminent nationwide grid collapse.

July 27 and August 3, 2026 — grid access shifts toward supply-or-curtail. PJM’s board directed filings under which new large loads that neither bring new generation nor otherwise secure supply would face curtailment during capacity shortages. Texas began auditing generation plans and ERCOT reliance days later. MindCast named secured electricity as the decisive competitive advantage in August 2025, in its first data center study.

MindCast published that first study on August 21, 2025 — before the opposition wave, before the first moratorium, before Gallup measured the public. Twelve months of dated claims now meet a public record anyone can check. The pages below set each claim against what happened and name an outside source for every event.

Forecasts, structural reads, frameworks, corrections, and open claims are graded separately in Section I.

I. How Foresight Predictions Are Graded

Grading requires a rule stated before the evidence. Five categories cover every entry, and each entry names its category at the top.

To maintain institutional modeling standards, the record separates forecasts from frameworks. A forecast names a time-bound, testable event and either occurs or does not. A framework states a structural rule that subsequent activity illustrates, evaluated by explanatory fit rather than forecast settlement.

Counts refer to the entries graded in this publication, not the complete MindCast corpus.

Several forecast entries carry more than one component, and components settle at different times. Where a claim has settled in part, the entry names the settled part and the open one.

II. Five Events of 2026, and What Bears on Them

Data center coverage this year has turned on a handful of events. Each traces back to a MindCast publication that addressed the underlying mechanism, and the category of each claim appears alongside it.

FERC orders all six FERC-jurisdictional grid operators to justify or reform large-load rules — June 18, 2026. The orders reach PJM, MISO, SPP, CAISO, ISO New England, and NYISO, and require each to address cost shifting and transmission cost transparency. Bears on The Federal-State AI Infrastructure Collision (January 2026). Checkpoint settled.

Texas halts approvals for data centers in ERCOT’s interconnection process — August 3, 2026, after fewer than ten percent of facilities reported power and water use. Bears on The Model AI Infrastructure Authorization Code (2026). Framework observed in practice.

PJM directs filings on a supply-or-curtail framework — July 27, 2026. New large loads that neither bring new generation nor otherwise secure supply by June 1, 2027 would face curtailment during capacity shortages, ahead of measures affecting other customers. Bears on The Bottleneck Hierarchy (August 2025). Structural read.

Puget Sound Energy proposes roughly five gigawatts of new gas — August 2026, drawing public objection from Washington legislative leaders. Bears on the physics-dominant branch of The Federal-State AI Infrastructure Collision(January 2026). Branch indicator supported; branch open through 2027.

Gallup finds roughly seven in ten Americans oppose local construction — May 13, 2026, with both Ohio gubernatorial nominees pledging moratoriums. Bears on Ship the Workload, Not the Power (September 2025). Constraint settled; relocation component open.

Four of the five events arrived within eleven months of the publication addressing them. Texas illustrates a scoring rule rather than confirming a prediction, and appears here because the rule predicted the failure that produced the pause.

III. Community Consent

Local resistance now shapes data center siting more than any other single variable. MindCast identified the mechanism a year before the current wave.

Ship the Workload, Not the Power (September 2025) — Forecast Entry with Settled Component

We said. Community trust would become the ultimate site selection filter, governing not merely whether a data center could be built but how quickly it scales and at what cost. Computing work would migrate toward places where power and public acceptance already existed.

What happened. Gallup reported on May 13, 2026 that roughly seven in ten Americans oppose construction of an AI data center in their area. Data Center Watch counted opposition groups rising from 396 to 833 across 49 states, with more than 300 state bills filed in the first six weeks of 2026. Both major-party nominees for governor of Ohio pledged a moratorium. In Washington, Cle Elum adopted an emergency moratorium three days after a $200 million project was announced.

Grade. Community consent settled as a governing constraint; workload migration remains open pending documented project movement.

The same study rated four regions. Northern Virginia was marked amber to red with advice to prepare to slow or relocate capacity, and the rating held. Eastern Washington was rated favorable, and Section IX records the correction.

IV. Which Constraints Are Real

Public argument about AI and the electricity grid has produced both documented shortages and considerable exaggeration. Separating the two carries practical consequences, because a developer, a utility commission, and a legislature all price risk differently depending on which constraints they believe.

The Power Stack (March 2026) — Structural Read with Supporting Evidence

We said. Transformer scarcity, transmission congestion in Northern Virginia and Phoenix, and cooling limits are documented constraints, each supported by procurement filings, published queue data, or operator disclosures. Claims of imminent nationwide grid destabilization are unsupported: utilities warning of collapse had not increased their own transmission spending, announced projects cited connection timelines contradicting their stated urgency, and weather volatility and industrial load shifts remained available as alternative explanations.

What happened. The McKinsey Global Institute reported on June 30, 2026 that lead times for generators, chillers, transformers, and switchgear have more than doubled since 2019, with some exceeding three years, and that average grid connection waits exceed four years and reach a decade in some markets. Both constraints named in March were confirmed by outside measurement within four months.

The Power Stack, on Strategic Behavior (March 2026) — Structural Read with Supporting Evidence

We said. Where computing demand outruns infrastructure by this margin, companies stop competing on price and compete instead to lock up scarce access — reserving positions in the interconnection queue, signing exclusive power contracts, buying land beside substations, and setting proprietary equipment standards that impose switching costs on later entrants.

What happened. FERC’s June 18, 2026 orders direct all six regional grid operators to address cost shifting, transmission cost transparency, co-location arrangements, and behind-the-meter generation. FERC’s orders confirm that queue administration, co-location, and behind-the-meter arrangements became formal regulatory issues within six months of publication. The exclusive-contract, land-strategy, and proprietary-standard components remain ungraded here.

Regulators and outside analysts confirmed both readings within four months of publication, on evidence the March study named in advance.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

V. Firm Power

Electricity supply has moved from an operating cost to a condition of service. Grid operators and state officials now ask whether a facility brings its own generation before asking anything else.

The Bottleneck Hierarchy in U.S. AI Data Centers (August 2025) — Structural Read with Supporting Evidence

We said. The AI infrastructure race would be decided by whoever secures reliable electricity, builds the right network connections, and designs cooling communities will accept — not by whoever buys the most computer chips. Firms holding secured firm power would outbuild rivals waiting on congested grids.

What happened. PJM’s board directed FERC filings on July 27, 2026 under which new large loads that neither bring new generation nor otherwise secure supply by June 1, 2027 would face curtailment during capacity shortages, ahead of Pre-Emergency Load Management affecting other customers. Large loads are not barred from connecting; they are served on a supply-or-curtail basis. Texas made self-supply an audit criterion in a directive now in force, and named companies committed to comply with the state’s standards, which require disclosure of generation plans, reduced ERCOT reliance, and water measures.

Grade. Strongly supported. The PJM board’s directive establishes supply-or-curtail as the operational threshold for large-load access, executing the mechanics described in August 2025. Final tariff approval rests with FERC.

VI. Federal and State Authority

Two governments claim authority over the same decisions, and the boundary between them determines what a developer must satisfy and who ultimately pays. MindCast has tracked the boundary since the Department of Energy first asserted jurisdiction in October 2025.

AI Computing Is Now Federal Infrastructure (November 16, 2025) — Forecast Entry with Settled Component

We said. Four consequences would follow the Department of Energy’s October 23, 2025 directive under Section 403: accelerated federal rulemaking on large-load interconnection, state regulators objecting that federal authority intrudes on their jurisdiction under the Federal Power Act, companies pursuing their own generation to bypass the queue, and transmission cost responsibility becoming the dominant economic fight.

What happened. Thirty-nine days later, on December 26, 2025, the Wall Street Journal documented state regulators invoking the 1935 Federal Power Act against federal oversight of large-load connections, litigation warnings from former FERC officials, and Florida moving to prohibit utilities from charging residents for hyperscale development.

Corporate self-generation is visible in Texas, where named companies committed to comply with the state’s data center standards requiring disclosure of generation plans and reduced ERCOT reliance.

The rulemaking chain ran as modeled. The October directive produced a FERC advance notice, roughly 200 stakeholder comments, and on June 18, 2026, six show cause orders under Section 206 of the Federal Power Act.

Grade. Three dynamics are documented. Corporate self-generation is supported by announced company plans and remains short of settlement.

One Settled Checkpoint and One Supported Branch Indicator from The Federal-State AI Infrastructure Collision (January 2, 2026) — Forecast Entry with Settled Component

We said. Nine checkpoints were published with dates and explicit failure conditions. Two are graded here. Cost allocation disputes would surface explicitly in PJM or MISO proceedings by late 2026, with dormancy stated as the failure condition. Separately, within the physics-dominant branch, natural gas would remain the dominant bridge fuel longer than public discussion acknowledged, on a horizon running through 2027.

What happened, on cost allocation. FERC’s June 18, 2026 orders require each regional operator to address cost-shifting concerns and provide greater transparency into how costs of serving large loads are identified, allocated, and communicated, with responses due August 17, 2026. Cost allocation became the explicit subject of a formal proceeding across all six FERC-jurisdictional regional grid operators, five months ahead of the checkpoint date.

What happened, on gas. Puget Sound Energy purchased six gas turbines and proposed roughly five gigawatts more, alongside converting the Centralia coal plant to gas, while Williams Companies announced Northwest pipeline expansion. Washington legislative leaders objected publicly on August 8, 2026 that the state’s clean electricity law was written to move off fossil fuels, not to build new gas plants.

Grade. Cost allocation settled five months early. The gas-bridge indicator is supported; the national branch remains open through 2027.

The Shape of Federal Authority — Structural Read with Supporting Evidence

We said. Federal authorities would win control over the process of connecting to the grid, while states retained authority over siting and over who pays.

What happened. FERC has proceeded through Section 206 tariff proceedings governing interconnection procedure, network upgrade cost recovery, and transmission service. Retail rates remain set by state commissions. H.R. 9419 creates a federal standard under the Public Utility Regulatory Policies Act and requires state regulators to consider and determine whether to adopt it, rather than imposing a retail rate directly — the federal-process, state-decision pattern the January analysis described.

Three Competing Governance Equilibria for AI Infrastructure — Framework Observed in Practice

We said. The contest resolves not along a single axis from permissive to restrictive, but into one of a small number of stable arrangements, and a jurisdiction can move on more than one dimension at once.

What happened. Texas paused approvals for data centers in ERCOT’s interconnection process on August 3, 2026 while continuing to court investment. President Trump called the pause a mistake and Governor Abbott defended it, with the Governor’s office citing an ERCOT-tracked surge in peak demand. Restriction and recruitment ran together inside one administration, under public federal pressure.

Federal and state authority now interact through formal proceedings rather than rhetoric, which makes the boundary observable and the next moves forecastable.

VII. Washington State

Washington concentrates the national question in one jurisdiction. Eastern counties host the load, western counties hold the legislative majority, and a 2019 clean electricity statute binds every utility in the state to a shared timetable.

Washington’s Clean Energy Advantage — Structural Read with Supporting Evidence

We said. Washington’s clean electricity requirements would function as a competitive asset or a competitive burden depending on institutional design.

What happened. Puget Sound Energy proposed roughly five gigawatts of new gas capacity in 2026, and legislative leadership objected publicly that the clean energy law was not written to produce new gas plants. Washington’s data center debate has become a fight over amending the statute, five months before the 2027 session convenes.

Under RCW 19.405, each utility carries obligations tied to its own retail load, so a large new customer in Grant County raises Grant PUD’s compliance burden directly. Cost reaches western Washington households through two indirect channels — competition for the same regional non-emitting resources, and carbon allowance pricing — and quantifying either requires evidence beyond the scope of this record.

New York’s Data Center Moratorium — Structural Read with Supporting Evidence

We said. A moratorium reaching only new applications and a moratorium reaching operating facilities are materially different instruments, and public debate routinely collapses the distinction.

What happened. The distinction became operative in Washington. Cle Elum’s emergency moratorium may not reach the project that prompted it, because the site sits under a development agreement the city approved in 2002 and the city’s attorneys are examining whether the older agreement controls.

Washington’s 2027 session will test both entries at once, since the state must resolve its clean energy obligations and its siting authority in the same drafting cycle.

VIII. What Permission Costs

Building a data center requires electricity, land, water, and capital. Before any of that, it requires permission — from a utility, a regulator, a county, and a state. MindCast treats permission as a good with a price, and the 2026 record has repeatedly demonstrated how that price gets set.

The Authorization Market (2026) — Framework Observed in Practice

We said. Permission to build, power, and operate a data center is scarce, unevenly distributed across states, and increasingly the binding constraint. Unresolved rules cost a developer more than strict ones, because a strict rule can be priced into a project and an unresolved one cannot.

What happened. Texas paused ERCOT-process approvals in August 2026 without a stated end date. Cle Elum adopted a moratorium while the legal effect of a 2002 agreement remains undetermined. Neither cost arises from the severity of a rule.

The Model AI Infrastructure Authorization Code (2026) — Framework Observed in Practice

We said. Any data center law can be scored on three things: what it obligates a company to do, what happens if the company does not comply, and how many projects it reaches. A requirement carrying no consequence scores low regardless of how strongly it is worded.

What happened. Texas asked data centers to report power and water use, and the Governor’s office reported that fewer than ten percent complied. On August 3, 2026 Governor Abbott halted approvals for data centers in ERCOT’s interconnection process and made the audit a condition of grid access, attaching a consequence to a request that previously carried none.

The Two-Ledger Data Center Bargain — Framework Observed in Practice

We said. Communities count what a data center brings — jobs, property tax, construction spending — and rarely count what it consumes: transmission capacity, water allocation, tax abatements, and grid headroom a future employer might have used.

What happened. Ohio’s Republican gubernatorial nominee proposed that data center companies cover the electricity bills of nearby residents. Texas’s August 2026 audit examines which facilities received state tax incentives, grants, or abatements. Both jurisdictions moved to make the second ledger visible, and neither has published a method for measuring it.

Permission pricing now drives outcomes in Texas, Ohio, and Washington simultaneously. The 50-State Baseline and the 50-State Regulatory Atlas score every jurisdiction on the same instrument, and Washington’s 2027 session will supply the first full test of the baseline’s central claim.

IX. Corrections

Three claims landed differently than written. Each appears below with the correction now carried forward.

Eastern Washington rated favorable, September 2025. Miss. The favorable regional rating did not hold after Cle Elum and Spokane adopted moratoriums in 2026. The rating scored host-county receptiveness and omitted the cost borne by ratepayers outside the host jurisdiction. Subsequent state-level modeling scores both variables.

Interconnection delays estimated at twelve to thirty-six months, September 2025. Miss on magnitude. The estimate materially understated delays now averaging more than four years and reaching a decade in some markets.

Texas forecast to accelerate, January 2026. Miss on tempo. The January analysis projected that Texas would accelerate recruitment as a beneficiary of federal standardization. Governor Abbott instead paused ERCOT-process approvals on August 3, 2026 while continuing to court investment. Future actor forecasts will test electoral-cycle proximity as a variable, since the scoring method carried no term for it.

Each correction now runs in the models that produce the 2027 state-level forecasts.

X. Open Foresight Predictions

Six open claims tracked in this record remain unresolved. Each carries the condition that settles it.

Antitrust attention migrates to the energy layer. Settles on a federal or state enforcement action, or a formal investigation, directed at interconnection access, exclusive supply, or queue position.

Patent leverage emerges in cooling and power equipment. Settles on a filed infringement action or a disclosed licensing dispute involving liquid cooling, power conversion, or grid-interface software at hyperscale.

Physics-dominant branch. Settles against the published 2027 horizon on whether dispatchable generation, equipment supply, and skilled labor fail to scale at the pace hyperscale demand implies.

Workload migration toward available power and consent. Settles on documented relocation or distribution of computing capacity away from constrained jurisdictions.

Washington closes the gap between enacted and proposed rules. Settles against legislation enacted in the 2027 regular session.

Hardware competition divides on infrastructure constraints. Settles on published market share at the 2030 horizon.

Appendix — The Publications

Every study referenced above, grouped by subject.

A. The Authorization Series (2026)

The Authorization Market: Standardized Bargaining, Rationed Power, and the Competition to Build America’s AI Infrastructure — Permission to build, power, and operate a data center treated as a good with a price.

The Data Center Authorization Price: A 50-State Baseline — All fifty states scored on the same instrument.

B. Standalone Reference Instruments

The Model AI Infrastructure Authorization Code — A scoring instrument grading any data center law on obligation, consequence, and reach.

The Data Center Authorization Market: A 50-State Regulatory Atlas — The full state-by-state regulatory record, maintained as a standing reference resource.

C. Governance and Public Benefit

Three Competing Governance Equilibria for AI Infrastructure — Three stable arrangements the contest can settle into, and what each requires to hold.

The Two-Ledger Data Center Bargain — What a data center brings measured against what it consumes.

New York’s Data Center Moratorium — Moratoriums reaching new applications compared with moratoriums reaching operating facilities.

D. Federal and State Authority

AI Computing Is Now Federal Infrastructure (November 2025) — Institutional response to the Department of Energy’s Section 403 directive.

The Federal-State AI Infrastructure Collision (January 2026) — Nine checkpoints with dates and explicit failure conditions.

AI Infrastructure Priority Under Scarcity — Who gets served first when power is insufficient, and who decides.

Why MindCast Is Filing a Public Comment with the Department of Energy — Entering the federal record on the 2026 National Transmission Needs Study.

E. Energy, Capital, and Competition

The Power Stack (March 2026) — Why shortages persist, and which constraints the evidence supports.

The AI Infrastructure Energy Opportunity Landscape — Investment concentrating in facilities while upstream shortages stay underfunded.

The AI Infrastructure Energy Antitrust Landscape — Competition enforcement moving toward control of electricity and grid access.

The AI Infrastructure Energy Patent Landscape — Licensing leverage arising once installed equipment makes redesign impossible.

F. Infrastructure Futures (2025)

The Bottleneck Hierarchy in U.S. AI Data Centers (August 2025) — Electricity, network topology, and cooling as the three governing constraints.

VRFB’s Role in AI Energy Infrastructure (August 2025) — Storage matched to twenty-year infrastructure horizons.

Nvidia’s Moat vs. Infrastructure-Customized Competitors (August 2025) — Infrastructure limits reshaping competition in AI hardware.

Ship the Workload, Not the Power (September 2025) — Community trust as the decisive site filter, with dated regional ratings.

G. Washington State

Washington’s Clean Energy Advantage — Clean electricity requirements as competitive asset or competitive burden.

H. Method

Predictive Institutional Cybernetics and Cybernetic Foundations — Institutions as decision systems that respond with measurable delay.

Constraint Geometry and Institutional Field Dynamics — Where physical and legal constraints allow development.

Chicago School Accelerated and Antitrust Enforcement Foundations — Why correction arrives after damage becomes structural.

Related works: AI Data Center Developers Are About to Disclose What State Power Approval Costs | The Authorization Market — Standardized Bargaining, Rationed Power, and the Competition to Build America’s AI Infrastructure | Three Competing Governance Equilibria for AI Infrastructure · Why MindCast Is Filing a Public Comment with the Department of Energy · New York’s Data Center Moratorium · The Two-Ledger Data Center Bargain · The Federal-State AI Infrastructure Collision · The AI Infrastructure Power Stack

Critical references: The Model AI Infrastructure Authorization Code · The Data Center Authorization Market: A 50-State Regulatory Atlas

I. Executive Summary 🎯

Every AI data center project in America now pays an authorization price — the accumulated legal and operational cost of permission to site, interconnect, power, and energize — and no two states charge it the same way. Twenty-four states have approved large-load tariffs. Nine carry dedicated enacted authorization floors — some statutory, some tariff-borne, some executive; the instrument type is recorded per state. More than one hundred localities hold binding pauses. The rest charge through silence, and silence is the most expensive price of all, because its terms get written after your capital commits.

Most readers arrive with one question — which states are strict — and the map answers it. But strictness is the wrong variable to shop on, and the paper’s core finding shows why: the strictest states keep winning the most bidders. Virginia demands $1.5 million per megawatt in collateral and holds the deepest data center market on earth. Texas mandates curtailment and drew 474 gigawatts of connection requests. Observed siting patterns are consistent with the reverse causal direction — capacity attracts load, and accumulated load induces stricter terms — and among legal variables, price certainty appears far more commercially decisive than nominal leniency, stated here at 75–85% interpretive confidence and tested directly in Publication 6.

Four findings anchor everything that follows. First, a Standard template already exists: the twenty-four tariff states converged on recognizably identical terms — 20–75 MW coverage, 85–90% minimum take, 10–15 year contracts — without coordinating. Second, the price moves in one direction: across the measured period, no state loosened an enacted term, and the year’s loudest restriction defeats removed proposed pauses without touching a single enacted one. Third, the real risk cells are not the strict states but the unwritten ones — California, Pennsylvania, Indiana, the tribal layer, and the accelerate belt — where terms arrive late, locally, and unbudgeted. Fourth, all of it is falsifiable: thirty-four predictions with deadlines and settlement sources let readers grade this paper in public through July 2028. Five headline the register.

Each carries a deadline, falsifier, and named settlement source in Section XI; misses publish at the same level as hits.

Who This Is For, and Where to Start

Three questions organize the paper for all four seats. Why do some states carry stricter laws and proposals than others? Sections 8 and 10 answer with a mechanism, not a mood: load causes law, grievances travel, and templates cascade. Why do hyperscalers choose some states over others? Section 8 prices the draw — speed-to-power first, certainty as the multiplier. How can a state optimize AI data center development? Section 12 answers with a playbook keyed to each state’s position, because the optimal move for Virginia is malpractice for Montana.

The document delivers the answers in reading order. Sections 2 through 5 set the architecture and deliver the fifty-state results, with the full evidence tables in The Data Center Authorization Market: A 50-State Regulatory Atlas. Sections 6 through 8 price the market. Sections 9 through 12 state the findings, run the forecast, freeze the register, and end where the audience acts: the optimization playbook by seat. The Data Center Authorization Market: A 50-State Regulatory Atlas — carries the full evidentiary layer for readers who want the receipts.

One sentence carries the thesis. The emerging American equilibrium is neither pause nor permission — it is authorization at a price, the price is knowable, and the firms and states that price it first will take the surplus from those that price it last.

II. How States Price Data Centers: Three Costs, Three Timing Strategies 🧭

Every data center creates three costs: energy (grid capacity, interconnection, stranded-asset risk), environment (water, emissions, land), and externalities on the public (ratepayer cost shifts, community disruption, the thin benefit line). No governance system disputes the costs. The three competing equilibria dispute only when the costs get priced along the build timescale:

The bargain is the unique time-consistent tempo, and the bargain is winning: the pause threat sets the floor on conditions, the federal acceleration layer sets the ceiling on price, and both threats price the middle without being exercised.

Each tempo maps to a layer. Acceleration lives federally because federal permits were never the modal bottleneck — states and counties decide roughly nine in ten approvals. Pause lives in proposal form federally and in execution form locally, with New York the lone state-level exception.

The bargain lives in the states because states own the pricing instruments: tariff authority, water permits, and the police power localities exercise on their behalf. As of mid-2026, the Edison Electric Institute counts 24 states with at least one approved large-load tariff and several more pending (EEI large-load tariff tracker, May–June 2026); an industry tracker counting proposed tariffs puts the total at 51 approved and 26 proposed across 36 states and 60 utilities (March 2026 update).

MultiState counted more than 300 data center bills filed across 30-plus states in the first six weeks of the 2026 sessions alone, after 200-plus bills in 40-plus states in 2025.

The map below scores every state on which of the three costs it has priced, using five measurement dimensions that roll up cleanly: cost causation and financial assurance price the energy cost; water conditions and disclosure price the environment cost; local consent prices the externalities cost (disclosure serves both of the latter — reporting is how the public sees the ledger).

Each dimension is scored strict-to-lenient with a trajectory arrow. Buyer-facing polarity flips the same data into operator vocabulary: friendly means terms are written and priceable; landmine means terms are unwritten and opposition is unpriced.

The strictest non-preempted layer controls the site, so the state row is a floor, never a ceiling — county and municipal action now moves faster than state action, and the local layer is where most 2026 surprises originated.

One structural finding leads everything else: New York’s EO 62 (July 14, 2026) is the sole state-level permitting pause in force at T₀ — scoped to DEC applications not yet deemed complete, expressly excluding local permits; Texas joined post-freeze with a broader audit-gated instrument (the Post-T₀ Addendum) — yet the moratorium threat has done its pricing work in at least six states without being broadly exercised. Maine passed one and the governor vetoed it into an advisory council. New York passed one and the governor substituted her own narrower executive pause. Georgia’s died in chamber and resurfaced as county ordinances.

The pattern is consistent with the framework’s core claim — moratoriums are negotiating postures that resolve into conditional statutes. Precisely stated at T₀: one completed conversion (South Dakota, pause tabled and disclosure statutes enacted inside a single session) plus three active or adjacent pathways (Maine’s advisory council, New York’s EO-gated framework, Georgia’s utility-and-county migration), tracked prospectively from this baseline via the instrument-change ledger.

One measurement bridge completes the architecture. The series umbrella publishes the Model Data Center Authorization Code — twelve provision families, three tiers, one strictness formula — as the instrument every series paper scores against, and the five dimensions below are the baseline’s reader-facing roll-up of those families: cost causation and financial assurance carry the tariff, queue, and assurance families; water and disclosure carry the environmental and transparency families; local consent carries the community and entry families.

Leaf-level Code scores live in the underlying dataset; the roll-up keeps fifty rows readable. And because strictness alone misleads, every row also carries the two buyer variables the Code records outside the strictness score — price certainty and authorization friction — so the four outputs a siting decision actually requires (strictness, certainty, friction, overlays) stay separately visible throughout.

III. How the Baseline Was Built — and What It Cannot Claim 🔬

Unit of analysis: state floor × utility territory × locality × tribal/federal nexus — authorization cost accumulates across layers while the slowest binding layer controls timing, so no single state color can carry legal status, stringency, and buyer consequence at once. Table 3a (The Data Center Authorization Market: A 50-State Regulatory Atlas) separates them: observed fields (Floor, Util, Local, T/F) record instruments and their status; inference fields (Tempo, price certainty, authorization friction, Trajectory) record MindCast’s classifications, labeled as such and banded.

Limitations, stated plainly. The utility overlay is one-third unmapped (❔ cells) pending the tariff-to-state attribution pass; most rows rest on secondary sources pending the primary-source backfill (The Data Center Authorization Market: A 50-State Regulatory Atlas G); trajectory arrows are assigned at a single baseline and validate nothing by themselves; and all throughput or count claims derive mechanically from the tables under the check protocol in the Verification section. July 30, 2026 is frozen as T₀: every forward claim lives in the prediction register against this baseline, and the instrument-change ledger (The Data Center Authorization Market: A 50-State Regulatory Atlas F) — empty at T₀ by construction — is the instrument that converts assigned arrows into observed movement.

One evidentiary rule governs the entire baseline and deserves plain statement. Anchor states — the jurisdictions carrying the paper’s named instruments — are primary-verified against commission orders, session laws, and executive texts. Remaining rows are desk-grade classifications built from the EEI survey, the DELTa tariff database, MultiState tracking, and trade coverage, published at the confidence stated per row and marked 🔶 or ❔ where the record is pending or unmapped. An ❔ cell is itself a finding — “no verified instrument located” — not a defect awaiting repair. Rows harden as they are touched: national conclusions rest on the verified spine and the aggregate record, while leaf-level certainty for any single jurisdiction is the engagement product, produced when a decision requires it. Every credible fifty-state survey works this way; the baseline differs only in saying so.

The Regional Service Overlay

PJM does not belong inside any state’s strictness score, and the omission is deliberate. The RTO crosses thirteen jurisdictions and imposes a different kind of price: the terms on which regional firmness, capacity treatment, and shortage protection become available. Under PJM’s July 27, 2026 proposal, a new 50 MW+ load that brings no capacity remains connected but is curtailed first during shortages — a service condition no state statute sets and no state score captures.

The baseline therefore distinguishes four cumulative price layers: the state authorization price (this document’s scores), the local overlay (the strictest operative county or municipal layer), the regional service overlay (RTO/ISO capacity, firmness, and curtailment terms), and the federal overlay (permitting, transmission, and needs-assessment policy). A Virginia project faces Virginia’s tariff and siting terms plus PJM’s capacity and firmness rules. One does not replace the other, and a siting decision that reads only the state row has priced two of four layers.

The utility-territory column in Table 3a (The Data Center Authorization Market: A 50-State Regulatory Atlas) records state-jurisdictional tariff terms and remains distinct from this layer. Publication 3 develops the connected-versus-firm mechanism in full; the baseline prices the layer’s existence, not its physics.

IV–V. The Fifty-State Results: 31 Bargaining States, One Pause, and a Local Veto Layer 🗺️

Full state-by-state evidence lives in The Data Center Authorization Market: A 50-State Regulatory Atlas — and the paper carries its results here so the argument stays self-contained. At T₀, the fifty states classify as 31 Bargain, 1 Pause, 5 Contested, and 13 Accelerate — the bargain posture already holds three-fifths of the map and a larger share of its committed load. Nine states carry dedicated enacted authorization floors, arriving variously as statute, commission order, approved tariff, or executive instrument; nine more hold partial instruments; twenty-four states have at least one approved large-load tariff; and more than one hundred localities hold binding pauses that set the consent floor beneath every state regime.

Two structural results from the map carry the rest of the paper. Pause barely exists at the state level — New York’s completeness-gated executive order is the only statewide permitting pause, and Delaware’s territory-scoped interconnection pause its only commission analog — while the pause tempo thrives locally, which locates the real veto power at the county line. And no verified binding jurisdiction moved toward unconditional acceleration during the measured period: every verified change wrote terms, tightened them, or conditioned a benefit. Tables 3a and 3b in The Data Center Authorization Market: A 50-State Regulatory Atlas record the per-state evidence, the reading key, and the three-cost pricing scores behind both results. One convention travels into this article: 🔶 marks a pending instrument or a placement awaiting primary verification, and ❔ marks a cell where no verified instrument has been located — an honest reading of the record, not a gap in it.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

VI. The National Template: How Twenty-Four States Wrote the Same Tariff Without Coordinating 📐

Convergence was the series umbrella’s boldest empirical claim — that a modal enacted template exists and legislatures are converging on it whether they know it or not — and the enacted record at T₀ confirms it. Confidence that the Standard tier is empirically real: 85–92%, consistent with TLSM-9 tracking at the top of its band.

Set the anchor instruments side by side and the template reads off the page. Ohio’s PUCO-approved AEP tariff: 25 MW class, 85% minimum take, 12-year terms. Virginia’s GS-5: 25 MW with auto-enrollment, 85% of transmission and 60% of generation demand, 14-year terms, $1.5 million per MW collateral. Oregon’s Schedule 96 under the POWER Act: 20 MW class, 90% minimum demand charges, 10-to-30-year terms scaling with load. Indiana’s I&M settlement: 80% minimum, 12-year terms. Texas, Georgia, Missouri, New Jersey: dedicated class or statutory mandate, cost causation, forecast credibility, contract durability.

The Code’s Standard tier — a 20–75 MW coverage class, 85–90% minimum take, 10–15 year terms, cost-causation language — is not a proposal. It is a description of what commissions and legislatures independently wrote between 2024 and 2026, and Atlas §XIII lays the full flexible-standard-strict rubric across every provision type.

Coase explains the mechanism and the speed. A standard contract is cheaper than fifty bespoke ones, and each approved instrument lowers the next jurisdiction’s drafting cost: commission staff cite prior dockets, legislative counsel copy enacted structures, utilities file terms their peers already litigated, and the White House ratepayer pledge concedes the cost-causation principle at the presidential level. §X names the dynamic as the template cascade, and the behavioral evidence that the terms bind is already measurable — AEP’s large-load forecast fell by roughly half after tariff approval, which is what a price does to speculative demand and what no incentive program has ever done.

Two qualifications bound the finding, and both matter for the 2027 sessions. The recurring financial terms converged; the frontier terms — per-customer clean procurement, verified flexibility, community-benefit mandates — remain the strict column’s province, and Michigan’s SB 1050 package is the migration indicator: passage pulls the modal template toward the strict column within two sessions (65–75%, per Atlas §XIII). And convergence is a floor phenomenon — the template standardizes the energy cost while leaving the environment and externalities costs to a patchwork, which is exactly where the local layer prices what states decline to.

VII. Why Strict States Are Safer Than Silent Ones: The Five Riskiest Cells on the Map 🧮

Commercial reality diverges from the strictness ranking the moment certainty and friction enter the model, and one comparison carries the whole argument. Virginia charges the highest visible authorization price in America: the strictest tariff stack of any major market, a first-in-the-nation consumption tax, Tier IV generator standards, cooling mandates, and a sixty-one-bill legislative docket. Virginia also cleared more committed data center capital in the measured period than any jurisdiction on earth, because every one of those terms is written, enacted, and datable — PC-H in Table 3a (The Data Center Authorization Market: A 50-State Regulatory Atlas) — and a capital committee can price a written term.

Indiana presents the inversion the PC/AF columns exist to expose. State-level terms sit at or below Standard and incentives run to fifty years, but nearly a third of Indiana’s counties enacted restrictions — eleven ordinances, seventeen-plus moratoria, outright bans in Marshall and Cass — making Indiana the widest state-local spread in the country: PC-L, AF-H. A developer comparing Virginia and Indiana on strictness alone chooses Indiana; a developer comparing them on the four outputs discovers that Indiana’s price is not lower, only later — and later is more expensive, because repricing after capital commits transfers the surplus to whoever writes the late term.

Silence performs worst of all, and the polling now quantifies why. Gallup’s March 2026 survey found 71% of Americans oppose a data center near them — 48% strongly — exceeding historical opposition to a nearby nuclear plant, and Data Center Watch counted at least 75 projects worth ≈$130 billion blocked or delayed in Q1 2026 alone, with opposition groups doubling to 833 across 49 states in a single quarter. A zero written price plus that opposition curve equals a consent term written mid-project by the least predictable author on the map: an organized locality with a portable grievance file. An uncapped exemption goes further still — under the Code’s classification rule, an unconditional incentive is a negative authorization price, a state paying firms to accept maximal repricing risk.

One more layer completes the silence argument, because the federal government is currently running the silence play at national scale. Expedited permitting does not lower the authorization price — it defers it. Executive Order 14318 expands categorical exclusions, narrows the federal actions that trigger environmental review, and channels qualifying projects through FAST-41’s accelerated schedules, which shortens the window in which local burdens surface before construction. Skipped review does not erase the burdens; it converts them into later, costlier forms — grievance, litigation, and consent price.

The Memphis-area record is the specimen: turbines installed at “warp speed” without permits in Southaven, Mississippi, and the bill arriving as a federal Clean Air Act suit, a $30–44 million annual health-damages estimate, and a grievance file now consolidated in the Harvard Law Review — the portable artifact that will be read into county hearings nationwide (Olivia Ferdinand, “Building at the Speed of AI: Data Centers, Expedited Permitting, and Who Bears the Burden,” Harvard Law Review Blog, Aug. 5, 2026). Borrowed speed is repaid at the consent layer, with interest — which is the deferred-price version of the rule this section keeps finding: the cheapest path at approval is routinely the most expensive path at settlement.

Five landmine cells dominate the T₀ map — cells where committed or arriving load is largest relative to the completeness of written terms. Indiana is the friction landmine above. California holds the largest unwritten cell in absolute terms: massive latent demand, a cost-shift study due January 1, 2027, the paired SB 886/887 instrument in committee, and no binding statewide framework in between — every project sited in the study window prices against terms that do not exist yet, in the state most likely to write strict ones.

Pennsylvania concentrates PJM exposure without a statewide framework: the regional overlay — capacity prices, the $13 billion in Q1 data-center-driven costs the PJM market monitor documented — is doing the pricing a statute has not, and a reader of the state row alone misses most of the bill (the Regional Service Overlay box, §III, states the four-layer rule). The tribal 🟨 tier — sixteen states with material tribal land base and no consultation framework — is unpriced by every commercial site-selection product on the market, and the Seminole Nation’s ban atop post-McGirt jurisdiction shows what an unpriced sovereignty term does when it fires.

The accelerate belt — thirteen states competing on incentives with essentially nothing written — carries the steepest grievance exposure per §X’s second loop: lowest written price, fastest-rising consent price, no instrument mediating between them, and prime habitat for the least authorization-equipped operator class.

One rule converts the landmine list into practice: never shop the map on strictness, and never read a state row without the overlays. The cheapest written cell with the least-written remainder is the most expensive cell on the board — and the county with an adopted ordinance beats the county with no rules, per the Atlas §XV playbook. Interpretive confidence: 78–86%.

VIII. What Actually Attracts Data Centers — and Why Strict Law Follows Them 🧲

A paradox sits in plain sight on the map above, and any reader shopping the strictness rankings will trip over it: the strictest states keep winning bidders. Virginia charges $1.5 million per MW in collateral and holds the deepest queue on earth. Texas mandates curtailment and drew 474 GW of connection requests. Ohio demands 85% minimum take and its cluster keeps growing. Under the naive model — strict law repels investment — none of that should happen. The section below dissolves the paradox rather than merely resolving it, and the dissolution starts with the correlation question, because the paradox only exists if strictness is treated as an independent variable. It is not.

The Correlation, Stated Precisely

Three variables need separating before any correlation claim means anything: hosting capacity (H) — the physical ceiling set by grid headroom, deliverable energy, transmission, water, land, and queue speed; attractiveness (A) — observed bidder behavior, measured by installed base plus committed pipeline; and strictness (S) — Binding Strictness per the Code. The T₀ record supports four correlation findings, each carrying its own confidence band.

Attractiveness and strictness correlate strongly and positively — because attractiveness causes strictness, not the reverse. Rank the states by installed base and committed pipeline, then rank them by Binding Strictness: the top of both lists is nearly the same list. Virginia, Texas, Ohio, Oregon, and Georgia are simultaneously the most-built and the strictest, spanning both parties, while the ideologically varied low-load states — Hawaii, Rhode Island, Vermont, Wyoming — cluster at lenient regardless of politics. The causal arrow runs A → S: load arrives, ratepayer pain accumulates, statutes follow within one to two sessions — the incumbency-stringency mechanism of Atlas §XXI, the template cascade of §X, and Prediction 10’s trigger chain, all describing the same lag.

Strictness is a lagging indicator of attractiveness, which is why reading it as a deterrent gets the sign of the relationship exactly backward. Interpretive confidence: 80–88%, consistent with TLSM-14.

Original capacity and attractiveness correlate strongly; remaining capacity and cumulative attractiveness correlate negatively. Capacity enables inflow, and inflow consumes capacity — so the incumbent magnets now hold the least remaining headroom precisely because they held the most original capacity. Washington is the terminal case: the hydro that built Quincy is fully subscribed, and Grant PUD’s 2,800-plus-MW queue prices scarcity the statute never wrote. The cross-sectional snapshot therefore shows strict, built-out states running short on headroom while lenient, empty states hold surplus — a pattern that superficially reads as “strictness kills capacity” and actually reads as “capacity attracted the load that produced the strictness and consumed the headroom.” Interpretive confidence: 82–90%.

Strictness shows near-zero independent effect on attractiveness once assets are controlled — and the four-cell matrix supports it. Capacity-rich lenient states (North Dakota, Wyoming, Montana) attract selectively: single power-anchored campuses, not clusters — power sells, leniency is incidental. Capacity-poor lenient states (most of New England) attract essentially nothing, which is the decisive cell: if leniency had independent pull, the cheapest regulatory postures in the country would show inflow, and they show none.

Capacity-rich strict states (Texas, until August 4) dominate the market. Capacity-consumed strict states split on exactly one variable — certainty: Virginia (strict, priced, knowable) keeps winning while Washington (lenient on paper, unknowable in fact) keeps losing, the eighty-point Binding-to-Frontier gap of Atlas §XVII doing the sorting. The residual effect of law on bidding that survives the controls is not the level of strictness; it is the certainty of it. Interpretive confidence: 75–85%; the formal elasticity test is Publication 6’s register, and this section states the hypothesis that register settles.

The correlation is tightening over time. Today’s capacity-rich lenient states are pre-ratchet, not differently governed: every accelerate-tier state that lands a first hyperscale campus files large-load legislation within two sessions (Prediction 10, 80–88%), so the A–S correlation strengthens mechanically as the frontier converts. A correlation produced by a one-way conversion process is not a coincidence to explain; it is the ratchet viewed cross-sectionally.

Why Strict States Keep Winning — the Four Mechanisms

With the correlation structure established, the paradox’s remaining content reduces to four mechanisms, each independently sufficient.

Demand is nearly price-inelastic at current authorization prices. Roughly $725 billion of 2026 hyperscaler capital expenditure must deploy somewhere, and the binding scarcity is capacity, not willingness to pay — so strict terms move the price of authorization without moving the quantity demanded. A siting committee comparing Virginia’s collateral against the cost of not building is not making a close call.

A written price is financeable; an unwritten one is not. $1.5 million per MW is a line item a capital committee models in an afternoon. An unpriced county revolt is a discount-rate problem no model prices well. Strict states have converted authorization risk into authorization cost, and capital pays a premium for the conversion — the certainty thesis of §VII, now with its mechanism named.

Strict terms are a moat for exactly the firms doing the bidding. Financial assurance rations by balance sheet, which prices out the speculative and neocloud tail — and the hyperscalers bidding into strict states are the firms the rationing favors. Ohio is the demonstration: the tariff halved the queue and the surviving projects got served faster. A hyperscaler’s revealed preference for Virginia’s terms is partly a preference for a queue its competitors’ phantom requests no longer clog.

Strict states are post-ratchet; lenient states are pre-ratchet. A developer signing a 12-to-14-year contract prices expected terms over the contract life, not opening terms. The strict state has had its political fight and settled it into statute; the lenient state’s fight is still coming, and it arrives after commitment, on the locality’s schedule, with the grievance floor higher than today. Today’s terms in a lenient state are the cheapest that will ever exist there — which is a warning, not an invitation. Interpretive confidence across the four mechanisms jointly: 78–86%.

The Draw Stack — What a Siting Committee Actually Prices

Nine variables, in the rough order a 2026 siting committee weighs them — and the ordering itself is the decade’s biggest change, because energy cost fell from first place to the middle of the stack while speed-to-power took its place. First, speed-to-power: on a $10-billion-plus campus, months of idle capital waiting for energization dwarf a two-cent-per-kWh differential, which is why fast-and-moderate beats cheap-and-queued everywhere it is offered.

Second, deliverable energy and capacity headroom — the min() of interconnection, deliverable energy, and accredited capacity that Publication 3 formalizes; annual-average abundance means nothing during a winter-storm hour. Third, transmission — the corridor either exists or the project waits a decade. Fourth, water — cooling collides with municipal and agricultural allocation, and the Eastern Virginia and Arizona records show water writing terms energy never did.

Fifth, fiber and latency, which the box below shows is workload-differentiated rather than universal. Sixth, land — contiguous, zoned, and politically survivable acreage. Seventh, construction workforce and trades capacity, the quiet constraint on parallel megaprojects. Eighth, fiscal terms — real money, but never load-bearing: no incentive has ever rescued a site the stack above rejected.

Ninth, certainty, which is less a variable than a multiplier on all eight — the four-output lesson of §VII restated as arithmetic.

Why the Workload Travels — and Why That Built a Fifty-State Market

A user in Bellevue running an AI chat session cannot tell whether the tokens came from Quincy or from upstate New York. Light in fiber puts the Bellevue-to-Quincy round trip at a few milliseconds and Bellevue-to-New-York at roughly 60–70 — paid once per request, against generation that takes seconds. For chat, batch, and training workloads, geography is economically irrelevant on the demand side, which is precisely what created a genuine fifty-state authorization market: the facility goes wherever power, water, and permission are, not where the users are. Two exceptions keep a metro premium alive: real-time voice and interactive inference, where tens of milliseconds are perceptible, and agentic workloads making dozens of round trips per task, where 60 ms compounds into seconds. Training is the pure case — placeless on the demand side, ferociously constrained on the supply side, needing contiguous gigawatts, cooling, and internal bandwidth rather than proximity to anyone. The map’s division of labor follows: training and batch capacity flows to the power frontier; latency-sensitive inference gravitates to the incumbent metros; and every state on the map is a candidate for something.

The Fifty States, by Draw Profile

Four draw profiles organize all fifty states, and the profile — not the strictness tier — is what a siting committee reads first.

Profile 1 — the incumbent magnets. Assets so deep the authorization price is a rounding error. Virginia holds the world’s densest fiber interchange, the largest installed base, the deepest operations workforce, and an ecosystem that reproduces itself — which is why it sustains the strictest financial terms in the country without losing a single major bidder.

Texas offers the only single-state grid: ERCOT’s freedom from FERC interconnection process made it the fastest speed-to-power jurisdiction in America, layered over all-of-the-above generation and unlimited land — the audit pause (Post-T₀ Addendum) suspends the speed advantage without touching the assets beneath it.

Georgia pairs the Atlanta fiber hub with Georgia Power’s build program — one of the largest utility capacity expansions in the country — and Southern Company’s appetite to serve. Ohio converted central-corridor latency, Great Lakes-adjacent water, and AEP headroom into the New Albany cluster, then wrote the national template tariff on top of it.

Arizona runs the Phoenix cluster on Palo Verde nuclear, cheap solar, and the chip-fab ecosystem — with water as the binding term the state prices instead of energy. Illinois holds the Chicago interconnection hub — the Midwest’s carrier hotel — under CEJA’s clean-scarcity overlay. Oregon is the hydro-legacy incumbent (The Dalles, Prineville, and the co-op-territory AWS corridor the POWER Act’s coverage hole exempts). North Carolina carries the legacy trio’s campuses (Google Lenoir, Apple Maiden, Meta Forest City) on Duke’s build program.

Profile 2 — the power frontier. Energy surplus, speed, and land sell; the buyer brings everything else. Iowa is the profile’s proof of concept — MidAmerican’s wind fleet drew Google, Meta, and Microsoft a decade before anyone wrote a data center statute. Louisiana and Mississippi are the profile’s current extreme: Meta’s $10B Richland Parish campus and AWS’s Madison County build ride Entergy’s gas-fired speed and special-session hospitality. Tennessee hosts the profile’s most contested specimen — xAI’s Memphis buildout on TVA power, with the air-permit fight showing the grievance floor arriving ahead of the statute.

Oklahoma (Google Pryor; gas plus wind, under the tribal overlay), Nebraska (public-power Omaha cluster), Kansas and Missouri (Evergy territory, the KC metro spanning both), New Mexico (Meta Los Lunas on solar), Utah (Operation Gigawatt recruitment atop Meta Eagle Mountain), Idaho (Idaho Power’s hydro mix carrying Meta Kuna and Micron), Nevada (the TRIC cluster — Switch, Google, Apple — now adding geothermal firmness through the Clean Transition Tariff), Wyoming (Cheyenne’s wind-and-legacy-coal cluster), North Dakota (Applied Digital Ellendale — wind, gas, and free cooling), Montana (the Great Falls 600 MW pipeline arriving ahead of the 2027 rate rules), West Virginia (the microgrid-district counter-model: bring-your-own-generation with preemption), South Dakota, Arkansas, Alabama(Meta Huntsville on TVA), and South Carolina (Google Berkeley County) fill out the tier.

Two hybrids straddle Profiles 1 and 2: Indiana, where the AWS New Carlisle, Google, Meta, and Microsoft commitments met the widest local revolt in the country, and Wisconsin, where Microsoft’s Mount Pleasant campus anchors a WEC build program.

Profile 3 — the constrained premium. Demand centers and clean-grid states where capacity is tight, terms are strict or forming, and the draw is proximity, water, climate, or carbon accounting. Washington is the profile’s cautionary tale — the original power-frontier state (Quincy’s hydro built the industry’s template) now fully subscribed and pricing worst on certainty. California holds the inference-proximity premium and almost nothing else a hyperscale training campus needs; its unwritten terms are §VII’s largest landmine.

New York pairs upstate hydro and the $19.4B Genesee County proposal with EO 62’s pause — the re-entry terms will define the profile’s east-coast price. Pennsylvania is the co-location frontier: the Susquehanna nuclear-adjacent Amazon deal and the Homer City redevelopment monetize PJM scarcity directly, under the regional overlay no state statute prices.

New Jersey (metro adjacency, new tariff), Maryland (NoVA spillover into Frederick, against county pauses), Minnesota(Meta Rosemount and Amazon Becker inside the carbon-free-2040 path and the nation’s first standalone water permit), Michigan (Great Lakes water and cooling climate, with the conditionality fight live), Colorado (Denver fiber, Xcel’s pending large-load filing), Delaware (PJM adjacency under the interconnection pause), and Florida (NextEra build speed and Latin-American latency, against hurricane exposure) complete the tier.

Profile 4 — the thin markets. The New England five behind ISO-NE’s cost and constraint (Connecticut, Massachusetts, New Hampshire, Rhode Island, and contested Vermont), Maine (whose veto fight was fought over a single Jay project — the exception proving how thin the pipeline is), Alaska, and Hawaii. Their lesson is the correlation section’s decisive cell: leniency without capacity attracts nothing, and several of these states are writing terms — pre-emptively, per Hawaii’s working group — for facilities that have not asked to come.

Anchor-evidence cells marked 🔶 rest on secondary or aging sources and re-verify per The Data Center Authorization Market: A 50-State Regulatory Atlas G before external citation; the table records draw structure, not scored strictness — the Atlas map and §XIII carry the scores.

Synthesis — Portability Creates the Market, Headroom Rations It, Certainty Allocates It

Three forces assemble the answer to why the map looks the way it does. Workload portability created a genuine fifty-state market — the demand side is nearly placeless, so supply-side conditions decide everything. Physical headroom rations which states can host at scale — the H variable, depletable and unevenly endowed, sets the feasible set. And certainty allocates demand within the feasible set — which is where law enters, as a multiplier rather than a gate. The strongest single piece of evidence for the entire framework is the paradox this section opened with, now inverted into proof: firms free to arbitrage all fifty states, facing near-zero demand-side switching costs, still pay Virginia’s price. Nothing about a market clears that verdict more decisively, and Publication 6’s elasticity register will grade it in gigawatts.

IX. Six Findings That Hold Across All Fifty States 💡

Financial assurance is capacity rationing. Collateral and non-refundable fees ($1.5M/MW Virginia, $50k/MW in the Texas draft) allocate scarce headroom by balance sheet, sorting serious operators from speculative queue positions without a single denial — AEP’s load forecast halved on tariff approval alone. Full mechanism: the Competition Layer section.

Queue position is property. Grandfather clauses, tariff vintages, and interconnection slots appreciate with every tightening behind them; Batch Zero deadlines and GS-5 auto-enrollment dates are the race’s real calendar. Full mechanism: the Competition Layer section.

Buyer competition is the fourth force. The pause threat floors the terms and the federal bypass ceilings the price; firm-versus-firm racing sets where the price clears inside the band — and it clears high, because a state facing a queue of bidders extracts what no single-bidder state could. Full mechanism: the Competition Layer section.

Local silence is riskier than a written ordinance. Moratoria cluster where local law has nothing to say; a county with an adopted ordinance is a priced negotiation, a county without one is an unpriced freeze waiting for a trigger. Full mechanism: the contested-tier playbook.

Community tolerance is a depletable commons. One operator’s conduct reprices every operator’s sites; the burned commons is now measured quarterly — Data Center Watch counted at least 75 projects worth ≈$130 billion blocked or delayed in Q1 2026 alone, roughly matching all of 2025, atop DLA Piper’s earlier $64 billion cumulative estimate, and enforceable, loss-preventing instruments delivered pre-coalition are the only refill mechanism the record shows working. Full mechanism: the Competition Layer and Moratorium Layer sections.

Certainty outsells permission. Strictness enters a developer’s model twice, with opposite signs — once as cost, once as certainty — and the record already sorts on the second: Virginia carries the strictest written stack in the country and the deepest committed capital, while the accelerate tier carries the cheapest paper terms and the fastest-rising unpriced consent cost.

Developers can finance a demanding but knowable authorization price more readily than unresolved local veto risk — the series’ commercial thesis, tested empirically in Publication 6, and already carrying its strongest specimen set: every enacted cohort is still winning projects, including the strictest (Virginia under GS-5, Texas under SB 6, Oregon under the POWER Act), while the widest Binding-to-Frontier certainty gap in the dataset belongs to Washington, the market pricing worst relative to its resources. Interpretive confidence: 75–85%. Full mechanism: §VII, §VIII, and Atlas §XVII.

X. The Two Feedback Loops That Will Write the 2027 Map 🔮

Forecasting the 2027 map requires modeling the institutions that will write it, so the trajectory analysis instantiates five cognitive digital twins and couples them: the Statehouse Twin (majority coalition, fiscal committee, election clock, affordability salience), the Commission Twin (docket precedent, ratepayer-protection mandate, staff capacity), the Utility Twin (load-forecast credibility, cost-recovery incentive, peer-tariff visibility), the Locality Twin (grievance file, consent leverage, moratorium authority, expiration calendar), and the Developer Twin (capital clock, certainty preference, queue vintage, portfolio optionality).

Consistent with MindCast disclosure practice, the section reports mechanisms, trajectories, and confidence classes; no formal simulation run, internal thresholds, or weights are claimed or disclosed, and The Data Center Authorization Market: A 50-State Regulatory Atlas H states the methodological posture.

Two feedback loops govern the coupled system, both running in the tightening direction — which is why the ratchet is a structure, not a streak.

The template cascade operates among the Commission, Utility, and Statehouse twins and is the causal engine beneath TLSM-9’s standardization. Every approved tariff lowers the next jurisdiction’s drafting cost: staff cite the prior docket, counsel copy the enacted structure, the utility files terms its peers already litigated, and the political cost of writing terms falls with each enactment because the template arrives pre-legitimized — the White House pledge and the PJM registry data flowing to thirteen statehouses are accelerants, handing legislators the numbers that make drafting effortless. The cascade explains the 2024–2026 spread from a handful of tariff states to twenty-four and predicts continued diffusion at declining marginal friction. Interpretive confidence: 80–88%.

The grievance floor operates between the Locality and Statehouse twins and extends TLSM-11’s commons mechanism. Local grievances travel: a disclosure shock in Seattle, a noise fight in Eagan, a rate increase in the PJM zone accumulate into a shared national file every subsequent hearing opens with, so the consent price rises everywhere at once regardless of any individual project’s conduct — opposition groups doubling from 396 to 833 in one quarter measures the loop’s gain directly. The floor rises monotonically and transmits upward: statehouses metabolize local pressure into statewide terms, which is the channeling loop’s state-level rendering — the pause tempo’s local victories becoming the bargain tempo’s provisions. Interpretive confidence: 75–85%.

Run forward, the coupled system produces the trajectory calls the register freezes. New York sits at the sharpest branch point, already half-resolved: the Statehouse Twin passed the pause, the executive substituted EO 62’s narrower completeness-gated version, and the modal path converts the pause into terms — the GEIS rulemaking writes the re-entry conditions while the Act’s delivery deadline forces the veto-or-amend decision by mid-January (Predictions 15 and 16).

Washington’s clock compresses: the workgroup report must land within roughly ninety days for the interim drafting window to function, and the successor runs as an energy bill with tariff machinery (Predictions 13 and 14; Atlas §XVII). Michigan is the Standard tier’s migration test running live in both directions — CES repeal pressure against the SB 1047–1051 conditionality package — and its 2027 disposition, reset by the November election, signals whether per-customer clean procurement and CBA mandates enter the national template (Predictions 18 and 19; Atlas §XVI).

California’s study clock expires in January with the paired-instrument negotiation open — the highest-probability large entrant to the bargain-enacted tier (Prediction 20). Virginia’s consumption tax sunsets into a successor fight the Statehouse Twin resolves toward continuation, because a revenue stream attached to an unpopular payer does not die in an affordability cycle (Prediction 21). Texas lands its five-track rule stack by December, converting draft strictness into vintage terms and starting the 2027 ratchet from a higher floor.

And the incentive-clawback cohort — Arizona, Illinois, North Carolina — is half-finished by construction: withdrawal without tariffs files cost-causation bills within two sessions (Prediction 10 logic, 70–80%).

Mechanism before outcome, then the outcomes: §XI states each trajectory as a settled-or-missed event with a date.

XI. The Predictions: Thirty-Four Falsifiable Calls Through 2028 📡

The register is the paper’s accountability instrument: thirty-four event probabilities, each settled through a named source, eligible for Brier scoring, graded in public with misses published at the same level as hits. Interpretive confidence in mechanisms appears in the body text and never pools with these numbers. Bands, deadlines, triggers, falsifiers, and settlement sources freeze together; revisions preserve the prior entry.

Two tiers organize the register, and explanatory weight — not probability — assigns them. Primary Predictions (1–12) test the system’s mechanisms: the ratchet, the template cascade, the sorting instruments, the certainty thesis. Secondary Predictions (13–34) test named jurisdictions and narrower classes, grouped by state so adjacent entries read together. A high-probability call about one chamber’s calendar stays secondary; a 70% call about ten siting decisions is primary, because it grades the framework itself.

Three freeze vintages apply: the original entries froze with the baseline on July 30, 2026, the Texas entry froze August 4, and the run additions froze August 6. Every entry’s full contract — trigger chain, falsifier, settlement source, denominator rules, and freeze date — lives in The Data Center Authorization Market: A 50-State Regulatory Atlas J, alongside a concordance to the identifiers used in the T₀ release. The tables below carry what a reader grades on: the claim, the band, the deadline.

Primary Predictions — Mechanism Tests

Secondary Predictions — Jurisdiction and Class Tests

The settlement calendar front-loads, which is the register’s design working as intended: Prediction 14 grades on October 31, Prediction 19 on November 3, Prediction 22 on January 1, and Prediction 16 by mid-January — four calibration checkpoints inside six months, before the July 2027 wave (Predictions 1, 4, 17, 24, 26, 34) and the July 2028 close-out grade the structural claims. Readers can grade every entry against its named settlement source without MindCast’s cooperation.

XII. What Each State, Developer, and County Should Do Next 🧭

Three questions brought this paper’s four audiences to the table, and the sections above have now assembled every component of the answers. Government-affairs teams want to know why the map looks the way it does and where to commit. Lawmakers — state, federal, and county — want to know what to write. The section below states each answer plainly, then converts the third into a playbook, because “optimize AI data center development” means something different in each of the four positions a state can occupy.

Question One — Why Do Some States Have Stricter Laws and Proposals Than Others?

Strictness follows load, with a one-to-two-session lag. States write data center law after data centers arrive, because arrival produces the three pressures statutes answer: ratepayer cost-shift pain, physical-resource collision, and organized local grievance. Virginia, Ohio, Texas, and Oregon are the strictest states because they are the most-built states — the incumbency mechanism §VIII establishes and Atlas §XXI names. Ideology barely registers: the strict tier spans both parties, and the lenient tier does too.

Proposal stringency follows a second, faster driver: transmitted grievance. New York, Washington, Michigan, and California carry the map’s most aggressive proposals not because they host the most load but because they combine organized opposition, affordability politics in an election cycle, and clean-energy statutes that new load stresses — the grievance floor of §X arriving in bill form ahead of the buildout. A stringent proposal is a leading indicator of local pressure; a stringent law is a lagging indicator of load. Both signal market maturity, and neither signals hostility — which is why reading strictness as repellent misprices the map.

Question Two — Why Do Hyperscalers Choose Some States Over Others?

Physical assets pick the feasible set, and certainty allocates demand within it. The §VIII draw stack orders the variables the way a siting committee does: speed-to-power first, then deliverable energy and headroom, transmission, water, fiber, land, workforce — with fiscal terms near the bottom and certainty as the multiplier on everything. Strictness never appears on the list, because the record shows it has no independent repellent effect: firms free to arbitrage all fifty states still pay Virginia’s price, and the cheapest regulatory postures in the country attract nothing without power behind them.

One sentence serves a government-affairs reader as the whole answer: developers buy megawatts, months, and certainty — in that order — and the state row is only a floor beneath the local, regional, and federal overlays that complete the bill.

Question Three — How Can a State Optimize AI Data Center Development?

Optimization is position-dependent, and the four draw profiles of §VIII define the positions. One principle governs all four: publish the price. Every playbook below is a variation on converting unwritten risk into written terms, because the certainty premium is the only lever a legislature controls that moves capital.

The incumbent magnets (Virginia, Texas, Georgia, Ohio, Arizona, Illinois, Oregon, North Carolina) optimize by defending certainty. The load is committed; the risk is repricing shock. The move is Virginia’s: strict, published, stable terms.

Pricing discipline defines the play — price the externalities cost deliberately before the grievance floor prices it chaotically, keep amendments prospective so existing contracts hold, and resist both sudden repricing and sudden generosity. An incumbent’s most valuable regulatory asset is a price nobody has to guess.

The power frontier (the accelerate tier, Iowa to Mississippi) optimizes by enacting the Standard bargain before the first campus, not after. Every frontier state is pre-ratchet: today’s leverage is the highest it will ever be, and terms written at recruitment beat terms written at grievance. The playbook is concrete — adopt the Standard column (20–75 MW class, 85–90% take, 10–15 year terms, cost causation), screen the queue for verified seriousness before phantom requests inflate it, convert uncapped exemptions into conditioned ones, and write the water and disclosure terms while the developer is still courting.

A frontier state that waits for its first fight writes worse law with less leverage two sessions later — Prediction 10 is the register entry that grades this claim.

The constrained-premium states (Washington, California, New York, and peers) optimize by manufacturing headroom and ending irresolution. Demand is not the problem; capacity and uncertainty are. The moves are speed-to-power reforms — interconnection acceleration, co-location and bring-your-own-generation pathways, storage and firmness rules that create servable megawatts — paired with the Washington lesson stated as a rule: an unpassed framework is worse than a strict one, because developers can finance a demanding price and cannot finance an unknowable one. Pass the bargain modularly, publish the re-entry terms, and let the pause convert into conditions rather than persist as fog.

The thin markets optimize by preserving options, not pre-fighting arrivals. Disclosure frameworks and enabling authority cost nothing and position the state for the moment power economics change; moratoria against facilities that have not asked to come spend political capital on nothing.

Two audiences sit outside the state playbooks and get one paragraph each. County councils hold more pricing power than most statutes do, and the record’s lesson is direct.

A written ordinance beats silence for both sides — adopted siting terms attract the operators who can meet them, an unpriced county attracts the ones who hoped no one would ask, and a moratorium works best as a bargaining instrument with a published conversion path, which is how most of the hundred-plus local pauses are actually resolving. Federal lawmakers control the thinnest layer on the map, and the equilibria analysis this series began with still governs: quantity instruments (moratoria, caps) lose to price instruments (terms, registries, cost causation) in every observed contest, and the federal layer’s highest-value contribution is the one PJM’s registry demonstrates — data that lets every other layer price accurately.

Reading the Baseline by Seat

Application also differs by chair, and the baseline was built for four of them. A governor’s office or legislative author reads the Atlas map and §XIII to locate the state against the nine floors and twenty-four tariff states, then drafts against the Standard column so every deviation is a choice rather than an omission. A government-affairs or site-selection team reads §VII and §VIII, prices the four outputs per candidate site, and scores the local, regional, and tribal overlays before trusting any state row. A commission or utility reads §VI and §X — the next docket will be easier than the last, and the templates arriving from peer jurisdictions are pre-litigated. An investor or strategist reads §XI and treats the register as the tradable version of the analysis: thirty-four dated claims that settle in public.

Customized work applies the same instrument at operational resolution. The five report families this publication powers — State Authorization Profiles, Multi-State Site Selection Comparisons, Site-Specific Authorization Stacks, Authorization-Trajectory Reports, and Governor’s Competitive Position Reports — resolve the decision the public baseline cannot: your state, your site, your session, your capital clock. Commissioning terms appear in the series umbrella; inquiries reach MindCast at mcai@mindcast-ai.com.

Three questions, three answers, one closing symmetry. Strictness follows load; capital follows certainty; and the state that understands both writes its terms early, publishes them plainly, and collects the premium the confused states pay.

🚨 Post-T₀ Addendum — Week One After the Freeze (August 3–5, 2026)

Four days after the baseline froze, the largest single instrument change in the dataset’s history arrived, and the discipline built into this document — a frozen T₀, a ledger that opens empty, arrows labeled as inference — is what lets the event be recorded instead of retrofitted.

The event. On Monday, August 3, 2026, Governor Abbott directed the PUCT and ERCOT to conduct a comprehensive verification and audit of data center projects advancing through the ERCOT interconnection queue — power consumption and on-site generation, water use and cooling, tax incentives received, ownership structures, and local-impact mitigation — with any project failing the audit to be denied grid connection (Abbott letter to PUCT/ERCOT, Aug. 3, 2026; Argus; KERA). ERCOT stated the directive “effectively pauses all data center projects” and postponed the Batch Zero transmission planning study (ERCOT statement via Fox News, Aug. 2026).

The directive is also the ERCOT-region analog of the regional service overlay taking shape in PJM: where PJM proposes to price firmness through capacity accreditation and curtailment priority, Texas now prices queue entry through verification — two RTO regions converging on the same principle that connection is earned, not assumed. The stated predicate is queue arithmetic the baseline’s competition layer anticipated: roughly 474 GW of connection requests — more than five times record peak load, up from 233 GW in January — with about 90% from data centers (Governor’s office; TechCrunch, Aug. 4, 2026). The directive builds on Abbott’s June full-cost-funding order and PUCT Chair Gleeson’s July 17 letter requesting broader legislative authority over the industry (ABC13, Aug. 4, 2026).

What it changes on the map. Texas’s current-state posture moves 🤝 BARGAIN → ⏸️ PAUSE (executive, audit-gated, ERCOT-region interconnections); the directive reaches projects advancing through the ERCOT interconnection process, not non-ERCOT territory, onsite-generation projects, or land-use permitting, so the map records an ERCOT interconnection pause rather than an unqualified statewide project pause.

Price certainty drops from H to L until audit standards publish; authorization friction rises from M to H; and the ⬆️ trajectory arrow assigned at T₀ converts from predictive inference to observed movement — the ledger’s first row. New York’s EO 62 is no longer the sole state-level pause in force: Texas now runs a broader one — EO 62 holds DEC permit applications in abeyance, while the Abbott directive holds the entire interconnection queue.

The “lock terms before the December rule stack” runway the Atlas §XIV Texas entry priced at roughly five months closed in five days, and the Batch Zero starting gun the Atlas §XX grandfather-clause analysis flagged for August 1 was pulled before it fired. Operators holding executed agreements and completed interconnections hold the appreciated asset the queue-position-is-property finding described; everyone behind them now holds an audit obligation.

What it validates, graded honestly. Direction first: the move is a tightening, and Prediction 1 — the ratchet — tests bargain→accelerate reversals, so the entry records the ratchet firing at maximum visibility rather than any strain on it. The incumbency-stringency read (Atlas §XXI; TLSM-14) gains its strongest specimen yet — the second-largest data center state, Republican-led, executing the strictest single instrument on the map. The executive-layer finding (Atlas §XV) gains its seventh gubernatorial instrument, with one refinement worth recording: the T₀ finding held that executives pull outcomes toward the bargain from whichever direction the legislature overshoots, and here the executive tightened ahead of the legislature — consistent with the affordability election cycle doing the overshooting instead.

The anti-speculation logic of Finding 1 scaled from tariff to queue: the audit is SB 6’s forecast-credibility principle extended to all 474 GW by letter, rationing headroom by verified seriousness rather than by collateral alone. And the grievance floor (§X) supplied the pressure the directive answered — local moratoria and bipartisan scrutiny across Texas, per the wire coverage, metabolized into state action exactly as the loop predicts.

What it does not mean. A pause pending audit is a pricing mechanism wearing a pause’s clothes — the same reading the baseline gave Delaware’s interconnection pause, now at fifty times the scale. The modal path is conversion, not permanence: the audit resolves into standing verification and disclosure requirements — a stricter Batch Zero — with compliant projects proceeding under codified terms, rather than into an extended blanket moratorium.

The Atlas §XIV registered call that Texas’s effective price rises again in the 2027 session (75–85%) moves to the top of its band, now pairing Abbott’s exemption-repeal priority with audit-derived conditions. Prediction 24 receives no progress credit yet, and register discipline is the reason: the entry requires commission-level pauses pending tariff development — the Delaware mechanism — and the Texas event is a gubernatorial audit directive, not a tariff-development order. The directive counts as adjacent supporting evidence for the mechanism’s spread; it qualifies under Prediction 24 only if the PUCT later issues a tariff- or rule-development order meeting the prediction terms. Grading a prediction after watching the event is exactly what the freeze exists to prevent.

Prediction 17 — August 4, 2026. Texas resumes data center interconnection approvals under codified verification requirements — audit-derived standing rules via PUCT/ERCOT protocol or 2027 session law — by July 31, 2027, rather than extending a blanket pause or reverting unchanged to the pre-audit process. Class: Event. Band: 72–82%. Falsifier: the pause extends past July 31, 2027 without standing rules, or approvals resume under the pre-audit process without new verification requirements. Settlement: PUCT orders, ERCOT protocols and planning-study record, Texas session law.

The polling floor, measured. One day earlier, national polling quantified the grievance floor’s gain: J.L. Partners’ July newsletter found 43% of registered voters believe the country already has too many AI data centers, up from 35% in January — eight points in six months — against 4% who say too few, with 23% unsure (Yellowhammer News, Aug. 3, 2026). Two findings inside the numbers matter more than the topline. A presidential endorsement moved local support only four points (23% to 27%) — celebrity acceleration barely dents the floor.

And the firm found little statistical relationship between a voter’s general opinion of AI and support for a nearby facility: voters price the local ledger — jobs, tax revenue, infrastructure, community impact — which is the Two-Ledger claim in survey form, and the reason enforceable local instruments purchase what national framing cannot. Alabama’s Cygnal numbers (April, GOP primary voters: 22% good, 25% bad, 37% mixed, 16% unsure) locate the bargain constituency precisely: the persuadable 53% in the middle is who the terms are written for.

Cle Elum, Washington — the county playbook running live. Two days after the Texas directive, the framework’s smallest-scale specimen arrived: three days after a developer unveiled a 20 MW, $200 million data center in Cle Elum — a town of two thousand in the Cascade foothills — the City Council unanimously passed an emergency six-month moratorium on new facilities, citing power, water, fire protection, noise, and the local economy (Seattle Times, Aug. 5, 2026). Silence, arrival, emergency terms: the exact sequencing §XII warns both sides about, executed in seventy-two hours.

Five of the baseline’s claims fire in one small-town story. The pause is built to convert, not persist — an advisory committee, a September 8 public hearing, and a six-month clock are an ordinance under construction, which is Atlas §XIX’s account of how the hundred-plus local pauses actually resolve. The quantity instrument’s limits were stated by its own author: a councilmember warned residents the moratorium was “not a magical Band-Aid” and might not override the site’s preexisting development agreement — which means the ban may not stop the only project it was aimed at, while vested rights do at the municipal layer what vintage terms do at the tariff layer (Atlas §XX).

The 20 MW threshold recurs yet again as the modal coverage number, this time organically. And the utility supplied the Standard tier’s core principle unprompted: Puget Sound Energy confirmed the project could be served “with the customer funding the infrastructure upgrades necessary to protect existing customers from cost increases” — cost causation as routine practice, in a state whose legislature has never required it.

Wenatchee Valley, Washington — the other face of the same local layer, and the information market underneath it.In the same week Cle Elum paused, the chamber of commerce ninety miles north hosted Microsoft, Sabey Data Centers, and the NCW Tech Alliance to a welcoming room, citing more than twenty years of coexistence with the industry in the hydro corridor (Wenatchee Valley Chamber of Commerce, Aug. 5, 2026). Washington’s statute is silent in both places, which makes the pairing the week’s cleanest demonstration: the local consent market sets the state’s real authorization price, in opposite directions simultaneously, and incumbency lowers the grievance floor the way no statute can — twenty years of demonstrated benefits is a consent asset the newest entrant cannot buy at any price. Chamber-circuit work by incumbents is bargaining capability in action, the kind Publication 2 scores.

Two details inside the panel carry independent weight. The chamber urged residents to seek information “directly from the source” — a line organizations use only when contested claims already circulate — and the misinformation now moving through professional networks as well as neighborhood ones acts as a subsidy to the grievance floor: a distorted water or rate figure travels the same portable-grievance channels as an accurate one and raises the consent price in jurisdictions the project never touched.

Disclosure is the hedge, and the market has noticed — verified public reporting is the only durable counter to viral claims, which is a structural reason the transparency family keeps spreading and a commercial reason certainty infrastructure and credibility infrastructure are converging into the same instrument. And the chamber marketed Washington’s “nation’s highest environmental and regulatory standards” as reassurance — the pro-development coalition citing strictness as legitimacy, which is the certainty thesis confirmed from the least expected direction and adjacent evidence for the governance-capability-marketing entry (Prediction 27).

One week past freeze, the system is behaving as modeled: the ratchet ratchets, the floor rises, the executive valve turns, a gateway town writes the local layer’s next ordinance in public while an incumbent corridor renews its consent over coffee, and the ledger — built empty on purpose — has its second row.

How the Baseline Stays Current

Publication does not end the analysis; it starts the clock. The baseline is a living, versioned instrument: v1.0 froze at T₀ = July 30, 2026, every subsequent instrument change appends a dated row to the change ledger rather than rewriting history, and the Post-T₀ Addendum model — analysis dated, priors preserved — governs all future updates. The 2027 legislative sessions carry a standing watch list, and updated baseline versions publish as the record moves.

Accountability runs on the same clock. The thirty-four predictions validate in public through July 2028, with the first checkpoints arriving as early as October 2026, misses published at the same level as hits, and readers can grade every entry against its named settlement source without MindCast’s cooperation. Commissioned work updates fastest of all: each State Authorization Profile or site engagement hardens its rows from desk grade to primary-verified, and the hardened rows flow back into the next public version. Write to mcai@mindcast-ai.com to commission, correct, or contribute to the record.

Publication Information

Publication 1 of 6 in the AI Infrastructure Authorization Series — Primary Article · MindCast AI, LLC · Data cutoff and baseline freeze: T₀ = July 30, 2026 · 50-State Baseline v1.0 · Post-T₀ Addendum through August 5, 2026 · v2.9 (August 6, 2026): register expanded to thirty-four entries under the plain-number scheme, section headers revised, full prediction contracts moved to The Data Center Authorization Market: A 50-State Regulatory Atlas.

Publication note: the paper above is Installment I’s primary article. The Data Center Authorization Market: A 50-State Regulatory Atlas — carries the full evidentiary layer: the fifty-state evidence and pricing tables (3a/3b) with their reading key, per-state detail, the contested-state playbook, the scoring machinery, and The Data Center Authorization Market: A 50-State Regulatory Atlas A–J including the instrument-change ledger, source register, and the full prediction contracts with the identifier concordance. Section numbers are continuous across the pair: §IV–V and §XIII–XXII appear in the Atlas.

Critical references: The Model AI Infrastructure Authorization Code · The Data Center Authorization Market: A 50-State Regulatory Atlas

Executive Summary

America’s AI buildout still depends on chips, capital, power, and demand. But authorization — the accumulated legal and operational permission to site, interconnect, power, and energize a data center — has become the binding marginal input for an increasing share of projects. States have responded less by regulating the industry in the traditional sense than by standardizing the bargain, producing a functional competition among firms to win it.

Authorization now trades in a market with prices, buyers, sellers, and a visibly one-directional trajectory — and no common map of that market exists. Twenty-seven states now write data center law in twenty-seven vocabularies, hyperscalers negotiate against terms no instrument grades, and the national innovation consequences of next year’s legislative sessions remain unmodeled.

MindCast AI is publishing a six-part series that maps the authorization market end to end. One publication prices the sell side across all fifty states. One rates the buy side firm by firm. One prices the energy substance the bargain trades in. One renders a single statehouse at vote-by-vote resolution. One publishes the measurement code every other paper runs on. One converts the whole system into a forecast of the national AI innovation frontier through 2029. Authorization accumulates across four layers — federal transmission and permitting policy, RTO/ISO market and reliability rules, state utility and authorization law, and local and tribal consent — and the series prices all four.

Every analytical publication that advances forecasts carries falsifiable predictions with deadlines, falsifiers, and named settlement sources; the Code is the exception by design — a measurement instrument proves itself through adoption and application, not through a validation register. Every prediction sits timestamped on the public record, beyond quiet revision, and the full register is validated against public outcomes on two dates — July 2028 and July 2029. Readers who disagree with the analysis do not need to argue with it in the abstract; they can wait for it to settle.

Two anchor publications precede the series and establish its foundations: Three Competing Governance Equilibria for AI Infrastructure, which argued that conditional acceleration wins as the coordination equilibrium, and the DOE National Transmission Needs Study Comment, which supplied the federal measurement layer. The publications below extend those anchors into a complete operating map — and the measurement instrument itself, the Model Data Center Authorization Code, publishes in full as Part Two of this document.

Six Findings That Organize the Series

Architecture can wait; the insights the series defends should not. Six findings organize everything below, and each one is falsifiable in a paper of its own.

1 · State data center laws look like restriction and function like market design (Publications 1 and 5). Legislatures, commissions, utilities, and local governments are standardizing recurring entry terms, reducing the transaction cost of bespoke negotiation and producing a functional competition for scarce authorization. Across twenty-four states with approved large-load tariffs, the convergence centers on recurring financial terms — dedicated classes, cost causation, minimum-payment obligations, collateral, extended contracts, and exit protection — while a smaller frontier group adds verified flexibility, environmental performance, and community obligations.

Coase explains the first move: a standard contract is cheaper than fifty bespoke ones. Demsetz explains the second: the result resembles competition for a scarce franchise substituting for regulation of the market, even where no jurisdiction conducts a literal auction. Chicago accelerated — the price-theory tradition applied at legislative speed — in statute form. Interpretive confidence in the market-design reading: 80–88%.

2 · Permission is rationed, and power sets the ration (Publication 3). Variable clean resources can supply a large share of annual data-center energy before they can provide accredited capacity or firm hourly service. The physical ration therefore comes from the combined scarcity of deliverable energy, generation capacity, transmission, storage, and reliability headroom — not from renewable generation’s annual contribution alone — and it bounds the volume of new large-load authorization any jurisdiction or utility territory can support over a planning horizon. The constraint is physical before it is political — a structural claim held at 90–95%.

3 · Slots clear to posture, and governance capability is the currency (Publication 2). The rationed slots go to the firms that can meet the standardized provisions — collateral, clean-portfolio depth, verified flexibility, conduct — which makes governance capability competitive capital rather than compliance cost. Governments discount every developer promise by the firm’s record: a performed agreement in one jurisdiction lowers the discount every other jurisdiction applies, and a breach raises it everywhere at once. Interpretive confidence in the cross-jurisdiction discount mechanism: 75–85%.

4 · Restrictive-but-priceable law outperforms lenient-but-uncertain regimes (Publication 6). Developers finance certainty, not permissiveness: a demanding state with knowable terms attracts capital that a permissive state with volatile local politics cannot hold. The claim is the working hypothesis of the series’ forecasting paper, held at interpretive confidence 70–80% pending its register.

5 · Connected is not firm (Publication 3). Interconnection, access to energy, and capacity-backed firm service are three different products, and on July 27, 2026 the board of the largest U.S. regional grid operator directed a FERC filing proposing to separate them. Firmness is earned — through capacity, bilateral contracts, or enforceable curtailment — and a facility that earned none of it stands first in line for curtailment. The three-product separation is definitional; that regional rules will increasingly enforce it is forecast at 70–80% (AIAS-2).

6 · The price compounds across four ledgers, and the ratchet runs one way (Publications 1 and 4). Federal policy, RTO/ISO rules, state law, and local-tribal consent each price a different product, and the obligations compound across all four — Section III states the controlling rule. Meanwhile the trajectory runs one way: no verified binding jurisdiction in the T₀ dataset has moved toward unconditional acceleration during the measured period — an observed fact at T₀; its continuation is a register entry in Publication 1, not an assumption here.

Six findings, six papers, one market model. The rest of this document shows how each finding gets tested, what each paper delivers, and what a reader in any seat can commission when the public version raises a private question.

The Series at a Glance

One table orients the whole set: each publication owns one primary function of the authorization market, even where multiple instruments apply to the same project.

Read down the table and the market assembles itself: sellers, buyers, the commodity, the statehouse where terms get written, the ruler that measures them, and the national consequence.

What This Publication Is — and Is Not

Readers should treat this piece as the map of the map: an orientation to the series’ architecture, its headline claims, and its accountability structure — plus the one component every other paper depends on, published in full. Part One frames the market, the method, and the six publications. Part Two is the Model Data Center Authorization Code itself, so that when the fifty-state baseline and its successors land, the instrument they score against is already on the record. Nothing here substitutes for the papers; each carries its own evidence base, prediction register, and settlement terms, and each stands alone for the reader whose seat requires only one of them.

Reading paths by seat. Governors’ offices and legislative authors: Findings 1 and 4, then Publications 1, 4, and 6. Government-affairs teams at hyperscalers and developers: Findings 3 and 5, then Publications 1, 2, and 3. Utilities and commissions: Findings 5 and 6, then Publications 3 and 5. Investors and strategists: Findings 2 and 4, then Publications 1 and 6. Every path ends at Section VI, where each seat’s customized reports are named.

I. Authorization Became a Market — and No One Mapped It

Authorization became the binding constraint on AI infrastructure quietly, jurisdiction by jurisdiction, without any single actor deciding it would. State legislatures enacted large-load frameworks. Utility commissions approved tariffs with collateral, minimum-payment, and exit-fee machinery. Localities imposed moratoria and community benefit terms. Federal overlays layered on top. Each instrument priced one slice of one project in one place, and no instrument priced the whole.

Practitioners feel the result daily. A government-affairs lead comparing two states compares incommensurable regimes. A governor’s office asking where its state sits against competitors gets anecdotes, not a table. A legislator drafting a data center bill starts from a blank page when twenty-four states have already enacted precedent. Capital committees approve billions against authorization assumptions no one has stress-tested.

The physical constraint underneath the legal one makes the gap urgent rather than academic. Variable clean resources can supply a large share of annual data-center energy long before they can supply accredited capacity or firm hourly service, so the ration comes from the combined scarcity of deliverable energy, capacity, transmission, storage, and reliability headroom. Each jurisdiction and utility territory can support only a bounded volume of new large-load authorization over a given planning horizon, and projects compete for that available grid, generation, water, and political headroom.

Slots clear to the strongest bargaining posture. Finding 1’s standardization is what makes the competition legible: because the recurring term sheet converges even as clean-energy, water, labor, local-authority, and enforcement terms still vary, the contest shifts from discovering the terms to winning under them. The open question is who wins under it, in which states, at what price, and the series exists to answer it before capital commits.

II. Who Is Writing This, and Why the Method Is Different

MindCast AI is a predictive behavioral economics and game theory simulation firm. We build cognitive digital twins — computational behavioral models of the institutions, companies, and stakeholders in a contest — and simulate their decision-making through scenarios of constraint: what each actor does when the fiscal committee balks, when tariff terms tighten, when a rival concedes first, when a moratorium bill changes the bargaining table without ever passing. The output is not commentary on what already happened. The output is falsifiable foresight about what happens next, with deadlines and settlement sources, validated in public.

Neither parent discipline is predictive on its own, and the apparent paradox deserves a direct answer. Behavioral economics describes how real institutions deviate from rational-actor assumptions; game theory maps the strategic structure of a contest; both remain descriptive until they run inside a simulation. Fudenberg and Levine reframed equilibrium as the outcome of a learning process rather than a static solution, and Fudenberg’s later work pressed the discipline on the predictive completeness of its models — whether a theory forecasts, not merely explains.

MindCast carries that lineage into applied institutional forecasting through three additions. Cognitive digital twins supply the actor-specific initial conditions — a given institution’s incentives, constraints, and installed decision patterns — the way meteorology’s equations forecast weather only once fed observed atmospheric state. Simulation under constraint scenarios traces trajectories rather than hunting a single equilibrium, because the interesting question in a statehouse is never where a fixed game rests but how the game itself gets rewritten as pressure rises.

The prediction register supplies the third addition, and it does what neither parent discipline ever imposed on itself: every forecast carries a deadline, a falsifier, and a named settlement source, and the misses publish at the same level as the hits.

The engine earns that discipline in public before it is applied to statehouses. MindCast stress-tests the same simulation stack in arenas with immediate, un-fudgeable settlement — predictions published and timestamped on the record, then validated against the outcome — precisely because a forecasting method that cannot survive fast objective validation has no business advising on slow consequential decisions.

Traditional government affairs runs on relationship; the series runs on geometry. Relationships tell an operator who will take the meeting. Geometry — the structure of incentives, constraints, coalitions, and clocks — tells the operator what the meeting can produce, and the second question is the one capital actually needs answered.

III. The Four-Layer Governance Stack

Authorization accumulates across four layers of government, and each layer prices a different product. Federal transmission and permitting policy sets the national pace and the terms of access to federal process and federal land. RTO and ISO market and reliability rules — the layer most site-selection analysis skips — govern what no state statute reaches: accredited capacity, regional resource adequacy, wholesale market participation, curtailment sequencing, transmission deliverability, and the difference between firm and conditional service. State utility and authorization law sets the retail price: tariffs, financial assurance, clean-energy terms, siting conditions. Local and tribal consent sets the terms on which a community says yes at all.

The economics still reduce to quantity versus price — Congress contests how much gets built, the layers below contest the terms — but the price accumulates across all applicable, non-preempted layers. Within each regulatory domain, the most restrictive operative term controls; across domains, the obligations compound. A county can block zoning, an RTO can condition firmness, a state commission can assign retail costs, and a federal agency can control federal-land access — no single layer controls the whole authorization.

The regional layer stopped being theoretical while this series was in production. On July 27, 2026, the board of PJM — the largest U.S. regional grid operator and the system serving Northern Virginia’s major data-center concentration — directed staff to advance an Interim Resource Adequacy Service proposal for FERC review. Under the proposed framework, a qualifying new large load that has otherwise interconnected but lacks sufficient capacity coverage by June 1, 2027 would receive conditional reliability treatment and priority curtailment during shortage periods, while a regional Large Load Registry would hand states zone-level reliability data for their own tariff and flexibility policy.

FERC has not approved the framework, and this series treats it accordingly. The proposal’s direction, however, confirms the premise the publications share: connection is not firmness, firmness is earned through capacity, contracts, or enforceable flexibility, and Publication 3 develops the mechanism in full.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

IV. The Six Publications

Each entry below follows one skimmable pattern: the paper’s sharpest insight first, its governing equation second, its scope third, the questions only the full paper answers fourth, and the customized reports it powers last. Unless otherwise stated, equation inputs represent normalized indices rather than raw legal, physical, or financial units. A reader can stop at any layer and leave with something usable.

1 · The Data Center Authorization Price — A 50-State Baseline

The insight: strictness alone is the wrong variable to shop on. Two states with identical headline strictness present opposite commercial realities — a demanding-but-knowable Virginia against a permissive-but-volatile Indiana — because the commercial decision requires at least four outputs — strictness, price certainty, authorization friction, and the applicable regional and local overlays — and the last three move independently of the first. The commercial thesis is that developers can finance a demanding but knowable authorization price more readily than unresolved local veto risk — held at interpretive confidence 75–85%, and tested empirically in Publication 6.

AP(site) = Σₔ Sₔ(site) Sₔ is the normalized score of the most restrictive operative term within regulatory domain d. Within a domain, the strictest non-preempted term controls; across domains, the scores add.

Every AI infrastructure project now faces a distinct authorization price accumulated across state law, utility terms, local consent, tribal exposure, and regional and federal overlays. The paper maps all fifty states across three costs — energy, environment, public externalities — and four tempos: accelerate, bargain, pause, contested. Nine states carry comprehensive enacted floors, twenty-four hold at least one approved large-load tariff, one hundred–plus localities hold the bargaining floor, and no verified binding jurisdiction in the T₀ dataset moved toward unconditional acceleration during the measured period.

The full paper answers: which cells of the map are landmines because the terms are unwritten; where authorization tightens next, on a frozen ledger; and — for a governor’s office or legislative author — exactly where your state sits against the nine floors and twenty-four tariffs already enacted, provision by provision.

Customized reports: State Authorization Profiles · Multi-State Site Comparisons · Site-Specific Authorization Stacks · Authorization-Trajectory Reports · Governor’s Competitive Position Reports

2 · Who Bargains Best with the States — A Field Guide to Authorization Bargaining Positions

The insight: speed without governance capital manufactures the very moratoria, service denials, and hearing delays it then suffers from — and the guide tests two propositions sharp enough to falsify. First, that the national market clears only several dozen new large-campus authorizations annually, against disclosed demand several times larger. Second, that a principal’s home jurisdiction applies the largest conduct discount, because local institutions possess the deepest firm-specific record. Both are stated as model outputs with definitions and falsifiers in the paper, not as established facts — the slot-clearing proposition at 70–80%, the home-state discount at 65–78%.

Slot Odds ∝ Capability Bundle × (1 − Conduct Discount) Capabilities add; conduct multiplies — and the discount updates on the firm’s record in every jurisdiction at once.

Roughly $725 billion in 2026 hyperscaler capex — the four-company estimate frozen at the series baseline of T₀ = July 30, 2026, with later reporting already moving the figure toward $745 billion — is chasing that constrained pool, and firms compete with very different endowments. The guide rates principals, neoclouds, developers, and capital across seven capabilities — collateral capacity, clean-portfolio depth, governance conduct, demand credibility, optionality, federal alignment, and speed. Under frameworks like PJM’s, the same capabilities now purchase firmness itself, not just goodwill.

The full paper answers: the firm-by-firm ratings against the public record; each principal’s easiest and hardest states by name; and — for a state screening applicants — whose demand is real before scarce headroom commits.

Customized reports: Bargaining Position Scorecards · Competitor Authorization Benchmarks · State-Specific Bargaining Strategies · Governance-Capital Repair Plans · Applicant Credibility Reviews for Government

3 · The Grid-Anchored Clean Power Bargain — Powering AI Data Centers Through 2040

The insight: a facility can procure 100% annual renewable energy and still be the first load curtailed in a winter storm — because annual matching, hourly carbon-free performance, and capacity-backed firm service are three different products, and most clean-power debate collapses them into one. A mandate written without the resource-adequacy layer produces facilities green on paper and curtailment-first in practice — a structural claim held at 88–94%.

Firm Service = min( Interconnection, Deliverable Energy, Accredited Capacity ) The weakest of the three products sets service quality — connected-is-not-firm in one function.

The governing question is never clean-versus-grid; it is whether new load finances enough incremental clean generation, firming, and transmission to preserve reliability without shifting costs onto existing customers. The paper stages the answer to physics: disclosure and cost causation immediately, 80% annual additionality near 2030, hourly targets maturing through 2035–2040, with PJM’s proposed resource-adequacy framework as the first major regional test of the mechanism.

The full paper answers: what a legislature can bind next session versus what must wait for the build-out; which commitments a developer can actually deliver on which dates; and the model state standard, drafted.

Customized reports: Project Power Compacts · Connected-Is-Not-Firm Assessments · Clean-Power Deliverability Reviews · 2030–2040 Compliance Pathways · Model State Power Standards

4 · One Vote Above the Floor — What Washington’s E2SHB 2515 Teaches Every State Writing a Data Center Law

The insight: the country’s broadest data-center framework cleared its principal policy gates but failed when fiscal design, implementation costs, and unresolved coalition trades converged in Senate Ways and Means — and the failure generalizes into a drafting rule any state can use: carry tariff, queue, reporting, reliability, and core performance duties in the authorization vehicle, and move severable fiscal instruments through coordinated companion legislation.

P(Enactment) = P(Policy Coalition) × P(Fiscal Clearance | Policy Coalition) Fiscal clearance depends on the package the policy coalition produces. Bundling drives the conditional term down; modular drafting raises it — and with it the product.

Washington passed E2SHB 2515 through one chamber by a one-vote constitutional cushion; the Senate companion set the strict ceiling; and unresolved fee, tax, implementation, labor, and agency-cost questions exposed the entire regulatory package to a late-stage fiscal veto. Washington’s 2027 session becomes the framework’s first live validation, with enactment of a comprehensive or near-comprehensive law by June 2027 forecast at 70–80%.

The full paper answers: the vote-by-vote architecture of how the coalition assembled and where it broke; which provisions traded as levers and which formed the immovable floor; and what the 2027 successor bill looks like — before the session opens.

Customized reports: Bill Architecture and Provision Audits · Legislative Coalition and Vote Maps · Provision Bargaining Matrices · Committee and Session Foresight Simulations · Governor’s Enactment Strategies

5 · The Model Data Center Authorization Code — Part Two of This Document

The insight: a statute, a tariff, an executive order, a county ordinance, and an RTO rule imposing the same term are the same term — and once every clause scores on coverage, obligation, and consequence, the “standard tier” turns out to be empirically real: a modal enacted template (20–75 MW class, 85–90% minimum take, 10–15 year terms, cost causation) that legislatures are converging on whether they know it or not. The Code doubles as a drift detector — when the modal template migrates, it registers before any single state’s news does. Confidence that the Standard tier is empirically real at T₀: 85–92%; its migration is forecast at AIAS-4.

Provision Strictness = 25 × ( 0.30 Coverage + 0.40 Obligation + 0.30 Consequence ) What the law requires outweighs how many facilities it reaches or how it punishes — obligation carries the largest weight by design.

Twenty-seven states are writing data center law in twenty-seven vocabularies — with more joining every session — and no common instrument grades them on one spectrum. The Code makes the model-code move for AI infrastructure: it strips away incompatible legal forms, preserves regulatory function, and scores every operative term on a common unit — the authorization price. Twelve provision families define the grammar of everything a data center law can say; each family carries lenient, standard, and restrictive tiers anchored to named enacted instruments; and statute, tariff, executive order, county ordinance, and RTO instrument all grade identically when they impose identical terms. The full Code follows below.

Part Two answers: what the modal enacted design looks like, so deviation becomes a choice rather than an accident; how a legislative office drafts against the standard tier; and how a GA team builds one national term sheet — varying parameters, not structures.

Customized reports: Authorization Code Profiles · Bill-to-Code Crosswalks · Model Bill and Tariff Drafting · Portfolio Compliance Matrices · Authorization Drift Monitor

6 · The Authorization Elasticity of AI Innovation — Forecasting the 2027–2029 Frontier

The insight: strictness enters a developer’s math twice, with opposite signs — once as cost, once as certainty — and the series’ working hypothesis is that the innovation-maximizing point on the spectrum is the standard tier, not the lenient one. If the hypothesis holds — interpretive confidence 70–80% pending the paper’s register — the legislature that enacts the modal template does more for national AI capacity than the one that waives everything.

Throughput(state) = Headroom × Certainty × Conversion × Energization — summed across statesStrictness enters twice with opposite signs, as cost and as certainty, and the working hypothesis holds ∂Throughput/∂Strictness positive through the Standard tier.

America’s AI trajectory through 2029 will be set less by chip supply than by what thirty-plus state legislatures do in 2027. The paper forecasts national capacity under three scenarios — standardization, fragmentation, bifurcation — scoring every session outcome against the Code and converting strictness, price certainty, and authorization friction into energized-gigawatt throughput, with a register keyed to primary observables: megawatts contracted and energized, queue-to-service conversion, announced-to-realized capex.

The full paper answers: which named 2027 sessions move national capacity most; the probability on each scenario; and — for federal audiences — the fragmentation case for performance standards, stated in gigawatts rather than rhetoric.

Customized reports: State Authorization Throughput Forecasts · Legislative Scenario Impact Reports · Portfolio Deployment Forecasts · Fragmentation and Bifurcation Reports · Governor’s Innovation-Capacity Reports

Six insights, six instruments, thirty reports — and one dependency structure that determines the release order, laid out next.

V. How the Set Fits Together, and When Each Lands

The publications divide one market into its working parts, and the Code (Part Two) is the shared instrument. Publication 1 scores the sell side against it; Publication 2 scores the buy side; Publication 3 supplies the energy substance the bargain trades in; Publication 4 shows the bargain being written in a live legislature; Publication 6 converts the whole system into a forecast of what it produces.

Release order follows the dependency. The umbrella and the Code launch together. The fifty-state baseline follows, then the field guide as its paired companion on the same frozen baseline, then the clean-power deep-dive, then the Washington case study after the November 2026 elections and before the 2027 session. The elasticity forecast closes the arc once the 2027 legislative calendar gives its branch points dates.

Accountability runs on one structure across the set. The series operates on a frozen baseline of T₀ = July 30, 2026, with a single instrument-change ledger; every forward claim sits timestamped on the public record — beyond quiet revision — with a confidence band, a deadline, a falsifier, and a named settlement source; event probabilities and interpretive confidence never pool; and the master register is validated against public outcomes on two dates — July 2028 and July 2029 — with misses published at the same level as hits. A prediction that cannot lose is not a prediction.

The umbrella stakes its own entries, and they cover what no single paper owns: series-level trajectories that cut across the publications. Paper-level predictions — Washington’s enactment odds, the ratchet, the standard-tier hypothesis — remain in their papers’ registers and are cited here, never restated.

Falsification is symmetric: fewer than 30 tariff states, no second RTO filing, no cross-layer statute, or a Standard tier that holds or loosens each validates as a miss at the same prominence as any hit.

VI. Customized Intelligence Built From Each Publication

Each public publication establishes a reusable analytical instrument. Customized reports apply that instrument to a named company, project, jurisdiction, bill, utility territory, legislative session, or capital decision. The client does not receive a generic restatement of the publication; the report resolves a specific decision — where to build, what terms to offer, which promises to trust, how a bill will move, what power commitments are credible, or how authorization conditions change expected deployment.

Where offerings touch the same project, they answer different decisions: Publication 1 selects and forecasts jurisdictions; Publication 4 models political passage; Publication 5 classifies and drafts legal instruments; Publication 6 forecasts deployment throughput. Eight products are easiest to conflate, so the table below states each one’s primary question before the full menu.

The six families below name every report and the decision it resolves — thirty products, five per publication, each one sentence long.

From Publication 1 — Jurisdiction Selection and Authorization Risk

State Authorization Profile Full Code-based assessment of one state, including strictness, certainty, friction, overlays, trajectory, and likely next move.

Multi-State Site Selection Comparison Selected jurisdictions compared across tariffs, financial assurance, power, environmental obligations, consent, incentives, and execution timing.

Site-Specific Authorization Stack A named site’s full obligation stack — state law, utility territory, RTO/ISO rules, county or municipal requirements, tribal exposure, and federal permits — with the most restrictive operative term identified in each domain.

Authorization-Trajectory Report Where a state’s authorization price moves over the next 12–36 months: pending bills, commission proceedings, utility filings, coalition pressures, and falsifiable branch predictions.

Governor’s Competitive Position Report How a state’s regime performs against selected competitors — which provisions improve project quality, which create avoidable friction, and where the state risks losing credible projects.

From Publication 2 — Firm Positioning and Counterparty Credibility

Authorization Bargaining Position Scorecard Confidential firm-level rating across collateral capacity, clean-power depth, governance conduct, demand credibility, geographic optionality, federal alignment, and execution speed.

Competitor Authorization Benchmark The client against selected hyperscalers, neoclouds, developers, utilities, or infrastructure funds competing for the same jurisdictional headroom.

State-Specific Bargaining Strategy What a named firm should offer, protect, trade, and refuse in a particular state — the concessions most likely to purchase certainty, speed, firm service, or political legitimacy.

Governance-Capital Repair Plan Which past conduct, unmet commitments, litigation positions, controversies, or disclosure gaps cause government actors to discount the firm’s promises — and which verifiable commitments repair the discount.

Applicant Credibility Review for Government Prepared for governors, utilities, commissions, or local governments assessing whether an applicant’s load forecast, financing, power plan, timeline, community commitments, and claimed flexibility deserve reliance.

From Publication 3 — Power, Firmness, and Performance

Project Power Compact A project-specific roadmap covering immediate cost-causation duties, resource procurement, annual additionality, hourly performance, storage, deliverability, flexibility, fallback supply, reporting, and enforcement milestones.

Connected-Is-Not-Firm Assessment Whether a proposed facility holds interconnection, deliverable energy, accredited capacity, and capacity-backed firm service — or remains exposed to shortage curtailment, market purchases, and contingent supply.

Clean-Power Deliverability and Maturity Review Proposed PPAs, onsite generation, batteries, advanced nuclear, geothermal, and transmission rights classified by operational maturity rather than announcement value.

2030–2040 Compliance Pathway A project or portfolio’s current power position translated into feasible annual and hourly milestones, with infrastructure dependencies, delay branches, and enforceable safety valves.

Model State Power Standard Customized legislative or regulatory language for a state, utility territory, or RTO region covering cost causation, resource adequacy, clean additionality, firmness, telemetry, curtailment, reporting, and remedies.

From Publication 4 — Legislative Strategy and Bill Simulation

Bill Architecture and Provision Audit Section-by-section analysis showing which provisions form the authorization floor, which belong in the bargaining zone, which create fiscal exposure, and which may operate as poison pills.

Legislative Coalition and Vote Map Chamber-specific analysis of sponsors, committee gates, swing members, utilities, labor, environmental groups, local governments, tribes, developers, and fiscal actors, with confidence bands around likely support.

Provision Bargaining Matrix A protect–trade–reject matrix for each operative clause, with acceptable fallbacks, counterpart interests, amendment risk, and the coalition effect of each concession.

Committee and Session Foresight Simulation CDT foresight simulation of a named bill through hearings, substitutions, fiscal referral, floor action, second-chamber changes, conference, and executive review.

Governor’s Enactment Strategy A coordinated legislative, agency, utility, fiscal, tribal, and public-message plan for converting an administration’s data-center objectives into a passable, implementable package.

From Publication 5 — Drafting, Classification, and Compliance

Authorization Code Profile A complete leaf-level score of a state, tariff, bill, executive order, RTO rule, or local ordinance across all twelve provision families.

Bill-to-Code Crosswalk Proposed legislation against the Lenient, Standard, and Restrictive tiers — omissions, deviations, internal inconsistencies, and enforcement gaps.

Model Bill or Tariff Drafting Customized legislative, commission, utility-tariff, or local-ordinance language built from selected Code positions rather than copied wholesale from another jurisdiction.

Portfolio Compliance Matrix Which existing and planned facilities satisfy current and expected obligations across jurisdictions, and where expansion, incentive, reporting, or service risks arise.

Authorization Drift Monitor A recurring report tracking movement in the empirical Standard tier as statutes, tariffs, commission orders, RTO rules, and local instruments change.

From Publication 6 — Throughput, Scenarios, and Policy Impact

State Authorization Throughput Forecast How much proposed load moves from announcement to contract, authorization, interconnection, and energization under a state’s present and expected regime.

Legislative Scenario Impact Report Alternative bills or amendments compared on how strictness, certainty, friction, incentives, and local contestability affect expected investment and energized capacity.

Portfolio Deployment Forecast How authorization conditions across selected states affect the timing, probability, and sequencing of a national project portfolio.

National Fragmentation and Bifurcation Report Whether state divergence produces standardized development corridors, delayed regions, or separate high-certainty and high-friction markets — and what federal intervention could change.

Governor’s Innovation-Capacity Report Whether a state’s proposed regime is likely to increase credible development, merely attract speculative announcements, or suppress throughput through uncertainty and cumulative veto risk.

Cross-Publication Flagship Dossiers

A few client questions require more than one instrument, and the combinations are named: the Market Entry Dossier(Publications 1 + 2 + 5) · the Project Authorization Dossier (1 + 3 + 5) · the State Authorization Legislative Strategy Dossier (1 + 4 + 5 + 6) · the Hyperscaler National Portfolio Strategy Dossier (1 + 2 + 3 + 6) · the Governor’s AI Infrastructure Compact (1 + 3 + 4 + 5 + 6) · and the Full Authorization Intelligence Dossier, drawing on all six.

The product structure mirrors the series exactly — sell side, buy side, energy substance, execution, measurement, payoff — so every report remains visibly derived from the public framework a reader can inspect for free.

VII. How to Commission Customized Work

The publications are the demonstration; engagements are the product. Every paper shows the engine’s outputs on the public record — validated predictions, the fifty-state map, the Code. An engagement points the same engine at your question, through the customized reports in Section VI — commissioned singly, or combined into the flagship dossiers and larger engagements. Standing work adds the Authorization Regime Map at operational resolution — live jurisdiction scores, trajectory calls, and the recommended move, before capital commits. Customized reports are scoped as fixed-fee projects. Standing monitoring, recurring score updates, legislative simulation, and portfolio-level advisory work are available through retainers with separately scoped project modules.

Noel Le · Founder, MindCast AI, LLC · mcai@mindcast-ai.com Publications: mindcast-ai.com · Simulation practice: mindcast-ai-simulation.com

Appendix 1 — Source Register

Time-sensitive and original claims in Part One rest on the following sources, frozen to the series baseline of T₀ = July 30, 2026 unless dated otherwise.

1. Twenty-four states with at least one approved large-load tariff — Edison Electric Institute, Large Customer Projects and Tariffs survey (July 2026).

2. Nine comprehensive enacted state floors; one hundred–plus verified local moratoria; the ratchet finding — MindCast T₀ dataset and instrument-change ledger underlying Publication 1 (fifty-state baseline, v1.0, T₀ = July 30, 2026), with per-count inclusion rules stated in that publication.

3. 2026 hyperscaler capital expenditure (~$725B at T₀) — four-company (Microsoft, Alphabet, Amazon, Meta) estimate as reported at the baseline date; Financial Times, 2026 hyperscaler capital-expenditure reporting (ft.com/content/dcf3873e-7b32-4a24-a90d-3bccf1d2c996) has since moved the estimate toward ~$745B, illustrating why the figure publishes frozen.

4. PJM resource-adequacy framework — PJM Board of Managers, Decisional Letter on CIFP Reliability Backstop Procurement and Connect and Manage (July 27, 2026); PJM Inside Lines, board-action summary; Utility Dive (July 28, 2026). FERC review pending; nothing herein assumes approval.

5. PJM governance proceeding — FERC Commission-Led Technical Conference, Docket No. AD26-7-000 (July 23, 2026); Statement of Principles Regarding PJM (January 15–16, 2026), signed by the National Energy Dominance Council chair and vice-chair and all thirteen PJM-state governors (energy.gov).

6. Washington legislative history — E2SHB 2515 (House passage 51–41; died at the fiscal gate) and SB 6171 (no floor vote), 2026 Regular Session records, Washington State Legislature.

7. Federal permitting and federal-land track — Executive Order 14318, Accelerating Federal Permitting of Data Center Infrastructure (July 23, 2025), 90 Fed. Reg. (July 28, 2025); DOE federal-site announcements (July 2025); CRS Report R48762.

8. Theoretical lineage — R. Coase, “The Nature of the Firm,” Economica 4 (1937): 386–405, and “The Problem of Social Cost,” Journal of Law and Economics 3 (1960): 1–44; H. Demsetz, “Why Regulate Utilities?,” Journal of Law and Economics 11 (1968): 55–65; D. Fudenberg & D. K. Levine, The Theory of Learning in Games (MIT Press, 1998); D. Fudenberg, J. Kleinberg, A. Liang & S. Mullainathan, “Measuring the Completeness of Economic Models,” Journal of Political Economy 130:4 (2022).

9. Authorization-slot throughput and home-state conduct discount — model outputs of Publication 2, published there with definitions (minimum campus size, unit of authorization, geographic coverage, duplicate removal, oversubscription estimation) and falsifiers; stated in this umbrella as hypotheses under test, not established facts.

Each citation above contains enough detail to verify the claim on its own, and the register updates only when the instrument-change ledger does.

Appendix 2 — Related MindCast Publications

The series extends a published corpus, and each prior work below supplies a specific framework the six publications build on. Every entry states its relevance so no reader needs to open it to follow this document.

• Three Competing Governance Equilibria for AI Infrastructure — the series’ primary anchor: conditional acceleration as the dominant coordination equilibrium, the nested governance model, governance capability as competitive capital, and the twelve-entry TGE register resolving through July 2028. Presented at the AI Infrastructure Summit — presentation brief.

• Public Comment — DOE 2026 National Transmission Needs Study — the federal measurement layer: net transmission burden during constrained hours, the seven-dimension commitment-quality screen, and three predictions in the federal docket resolving by July 2028.

• New York’s Data Center Moratorium — the infrastructure reclassification, the “charge them, don’t pause them” position the states have since adopted, and the AIRC-III register.

• The Two-Ledger Data Center Bargain — the behavioral mechanics beneath the bargain tempo: loss-aversion weighting, portable grievances, the Federal Acceleration Paradox, and the TLSM register.

• The Federal-State AI Infrastructure Collision — the partial-federalization forecast — federal control of process, state control of place and cost — now visible in the PJM governance proceeding’s state filing-rights debate.

• The Power Stack — the resource hierarchy behind deployment and the distinction between constraint-removal capital and choke-point capital, which Publication 3 extends into the firmness analysis.

• Predictive Game Theory Meets the Era of AI — the methodological foundation: operationalizing Fudenberg with cognitive digital twins, the game-replacement framework Section II summarizes.

• The Commerce Clause as America’s AI Advantage — the absorption-not-erasure preemption design informing how federal floors and state authority coexist across the four-layer stack.

The corpus and the series share one accountability structure: every register above is validated in public, on schedule, with misses published at the same level as hits.

PART TWO

The Model Data Center Authorization Code

The Standardized Provision Taxonomy, with Lenient-to-Restrictive Variations

Publication 5 of the AI Infrastructure Authorization Series · MindCast AI, LLC · Baseline freeze: T₀ = July 30, 2026

How to Use the Code

The document supplies one instrument for grading any state’s data center legislation along a single spectrum, in the tradition of model codes that translate incompatible state regimes into a common functional language: strip the legal form, keep the regulatory function, and score every clause on one invariant unit. Twelve provision families define the grammar of everything a data center law can say. Each family carries three tiers — Lenient, Standard, Restrictive — with the states currently occupying each band, and each leaf provision carries a canonical code so scores can always be reconstructed from raw statutory values.

Three companion uses follow directly. Legislative staff can draft against the Standard column and select deviations deliberately rather than by omission. Government-affairs teams can price a jurisdiction by reading its row positions. Analysts can grade fifty states on identical terms regardless of whether an obligation arrives as statute, tariff, executive order, or county ordinance.

§0 — The Normalization Rule: Score the Function, Not the Legal Form

Every provision, whatever its legal vehicle, is a price term in an authorization contract between the jurisdiction and the load. Identical price terms score identically whether they arrive as a statute, an approved utility tariff, an executive order, a commission rule, a binding local instrument, or an RTO/ISO instrument — tariffs, operating agreements, reliability rules, market rules, and FERC-approved service classifications all score by function, not vehicle. Form-blindness is the normalization: Ohio’s PUCO-approved tariff, Virginia’s enacted statute, and a proposed Michigan community-benefit mandate all reduce to the same three questions — who is covered, what must they do, and what happens if they fail.

Strictness measures the authorization price imposed on a project, not whether a policy is environmentally desirable or commercially predictable. Legal status, legislative tempo, price certainty, and authorization friction remain separate variables recorded outside the strictness score. A restrictive regime can offer high price certainty (Virginia); a lenient regime can offer none (a permissive state with unstable county politics). The code grades price; each state chooses its point on the spectrum.

The Standard tier is defined empirically, not normatively — the design the largest number of enacted instruments now share. Standard therefore migrates as the enacted record migrates, and the tier definitions below carry the T₀ freeze date for that reason.

§1 — The Grading Instrument

Each leaf provision receives three independent 0–4 scores. Raw thresholds, dates, percentages, and consequences remain in the underlying dataset so any score can be reconstructed and audited.

Provision Strictness = 25 × (30% Coverage + 40% Obligation + 30% Consequence), yielding a 0–100 scale per provision. Obligation carries the largest weight by design: a broad reporting rule should never automatically outrank a narrower quantified service condition.

Tier mapping. Lenient ≈ 0–33 · Standard ≈ 34–66 · Restrictive ≈ 67–100. The tier tables in §2 are the qualitative expression of the same bands, anchored to named instruments.

Aggregation rules.

• Binding Strictness scores only enacted statutes, effective executive orders, approved tariffs, and binding local instruments. Frontier Strictness scores active bills and proposed rules separately. Failed, expired, or vetoed measures receive no Binding or Frontier score and may appear only as labeled Design Precedents.

• Local Overlay records the strictest operative county or municipal layer for a candidate site. Authorization price accumulates across all applicable, non-preempted layers: within each regulatory domain the most restrictive operative term controls, and across domains the obligations compound — the state row is a floor, never a ceiling.

• A state’s Authorization Profile reports financial, clean-energy, environmental, community, and local-control strictness separately, so unlike regimes remain visible rather than collapsing into one misleading average.

• No incentive is neutral. A state scores as fiscally stricter only when it withdraws an existing subsidy, taxes consumption, or conditions a benefit on performance. An uncapped exemption is a negative price and scores Lenient.

One formula, three components, four aggregation rules — everything in the twelve families below reduces to them.

§2 — The Twelve Provision Families

Each family below states what it measures, lists its canonical leaf codes, and presents the three-tier ladder with the instruments anchoring each band at T₀. 🔶 marks a placement resting on pending rather than enacted language.

Family 1 — Scope and Coverage (SC)

Scope determines how much load, and how many facilities, enter the regime at all — the coverage denominator every other family multiplies against. A regime with restrictive obligations and a 100 MW trigger governs almost nothing; a 10 MW campus-aggregated trigger governs nearly everything.

Leaf codes: SC-1 load threshold · SC-2 campus / common-control aggregation · SC-3 new, expansion, and existing coverage · SC-4 data-center-specific vs. all-large-load scope · SC-5 geographic reach.

Family 2 — Entry, Siting, and Authorization (EN)

Entry provisions decide whether a project may enter ordinary permitting and obtain utility service at all — the family where a pause lives, and where the state–local allocation of veto power is written.

Leaf codes: EN-1 separate authorization requirement · EN-2 permit sequencing · EN-3 utility service discretion / refusal authority · EN-4 moratorium or pause · EN-5 state–local authority allocation · EN-6 federal land or facility authorization · EN-7 coordinated or expedited federal review.

Federal anchors: EO 14318 (July 23, 2025) — federal-land utilization, FAST-41 coordination, categorical exclusions; DOE’s four announced federal sites (binding federal instruments; site authorizations pending).

Family 3 — Queue and Interconnection (QU)

Queue provisions ration scarce grid headroom by proof of seriousness — the anti-speculation family. Phantom requests inflate forecasts, crowd out real projects, and stick ratepayers with study costs, which is why commercial-readiness screens spread fastest in the most constrained territories.

Leaf codes: QU-1 duplicate-request disclosure · QU-2 study deposits and fees · QU-3 site control · QU-4 commercial readiness · QU-5 attributable interconnection costs · QU-6 milestone loss and expiration.

Family 4 — Tariffs and Cost Causation (TF)

Tariff provisions answer the central distributive question: who pays the utility-system cost the new load creates. Cost-causation language is the single most standardized provision in the national record, and the family where bipartisan convergence is strongest.

Leaf codes: TF-1 separate tariff or rate class · TF-2 no-cross-subsidy rule · TF-3 full infrastructure cost recovery · TF-4 minimum demand charge · TF-5 generation vs. transmission allocation · TF-6 ramp schedule.

Family 5 — Contract Durability and Financial Assurance (FA)

Assurance provisions keep the customer financially responsible after long-lived infrastructure is committed — the stranded-cost family, and the one that rations authorization by balance sheet.

Leaf codes: FA-1 minimum contract term · FA-2 minimum billing / take · FA-3 collateral or letter of credit · FA-4 exit fee · FA-5 stranded-cost protection · FA-6 decommissioning security.

Family 6 — Energy Supply, Resource Adequacy, and Decarbonization (CE)

Energy-supply provisions determine whether new load adds clean and deliverable energy, secures sufficient resource adequacy, and qualifies for firm service — and on what clock. The family carries the sharpest analytical distinction in the code: annual matching proves procurement; hourly matching proves the power was there when the servers ran. A credible ladder imposes disclosure and additionality immediately while staging hourly performance through 2030–2040.

Leaf codes — sub-ladder CE-A, clean-energy performance: CE-1 clean share · CE-2 additionality (new resources only) · CE-3 compliance deadline · CE-4 annual matching · CE-5 hourly matching · CE-6 deliverability · CE-7 storage / clean-firm requirement · CE-8 carbon-market treatment (including carbon-allowance instruments such as Washington’s CCA treatment, design precedent). Leaf codes — sub-ladder CE-B, resource adequacy and firm-service qualification:CE-9 resource-adequacy coverage · CE-10 capacity-backed firm-service qualification.

A jurisdiction can be lenient on decarbonization and restrictive on firm-service qualification; the two sub-ladders grade independently and aggregate into one family score without hiding opposite policy positions.

Sub-ladder CE-A — clean-energy performance:

Sub-ladder CE-B — resource adequacy and firm-service qualification:

Family 7 — Reliability and Operations (RL)

Reliability provisions define how the facility supports or yields to the grid under stress — the family where flexibility stops being a marketing claim and becomes a service condition. Telemetry is the gateway leaf: nothing else in the family is verifiable without it.

Leaf codes: RL-1 telemetry · RL-2 flexible-load capability · RL-3 demand response · RL-4 emergency curtailment · RL-5 remote disconnection · RL-6 critical-load exemption · RL-7 backup-power standard · RL-8 shortage-curtailment priority.

Regional anchor (Frontier 🔶): PJM Interim Resource Adequacy Service — capacity-short large loads face priority curtailment in shortages, with proposed compensation for verified performance when directed to reduce, subject to FERC approval.

Family 8 — Water, Air, and Environment (EV)

Environmental provisions price the facility’s physical-resource footprint through permitting — water first, because cooling demand collides with municipal and agricultural allocation, and treaty resources where they run.

Leaf codes: EV-1 resource assessment · EV-2 water-availability finding · EV-3 dedicated water permit · EV-4 cooling-technology requirement · EV-5 generator emissions · EV-6 refrigerants and air · EV-7 land, heat, noise, traffic, wildlife · EV-8 treaty resources.

Family 9 — Transparency and Public Process (TR)

Transparency provisions determine what government and the public can verify without exposing protected facility details — the public-ledger family. Disclosure serves two costs at once: it prices the environment cost directly and the externalities cost indirectly, because reporting is how the public sees the ledger.

Leaf codes: TR-1 pre-development report · TR-2 annual operations report · TR-3 public aggregates · TR-4 granular agency access · TR-5 independent audit · TR-6 notice and hearing · TR-7 security / trade-secret safe harbor · TR-8 regional large-load registry.

Regional anchor (Frontier 🔶): PJM Large Load Registry — 50 MW+ at a single point of interconnection or within a one-mile radius; zone- and state-level reliability-impact data provided to states and distributors for flexibility and cost-allocation policy.

Family 10 — Community, Labor, Tribal, and Local (PB)

Community provisions allocate project benefits, voice, and construction standards — the consent family, where the local layer’s bargaining leverage becomes written obligation and where tribal exposure enters the code.

Leaf codes: PB-1 impact assessment · PB-2 community-benefit agreement · PB-3 host payment · PB-4 tribal consultation · PB-5 prevailing wage · PB-6 PLA / CWA · PB-7 apprenticeship · PB-8 local approval or rejection authority.

Family 11 — Taxes, Fees, and Incentives (FI)

Fiscal provisions decide whether public financial support remains unconditional or becomes reciprocal. The classification rule bears repeating here: no incentive is neutral, and a state scores stricter only when it withdraws, taxes, or conditions — an uncapped exemption is a negative authorization price.

Leaf codes: FI-1 consumption tax · FI-2 per-kWh public-benefit fee · FI-3 incentive duration · FI-4 performance conditions · FI-5 suspension or repeal · FI-6 revenue dedication · FI-7 clawback and sunset.

Family 12 — Enforcement and Remedies (ER)

Enforcement provisions determine whether failure produces a consequence proportionate to the obligation — the family that converts every other family from aspiration into price. An obligation without a remedy scores its Consequence component at zero, whatever its text promises.

Leaf codes: ER-1 corrective tariff or surcharge · ER-2 expansion hold · ER-3 permit suspension or denial · ER-4 service refusal or disconnection · ER-5 security draw · ER-6 tax clawback · ER-7 administrative penalty · ER-8 reconciliation and cure.

§3 — Reading Rules

Five rules govern application of the code, and each one prevents a specific misreading.

1. Form-blindness. Score the function, never the vehicle. A tariff, a statute, an EO, and a county ordinance imposing the same term receive the same score.

2. Standard is empirical. The Standard column records the modal enacted template at T₀ — the 20–75 MW class, 85–90% minimum take, 10–15 year terms, cost-causation language. It migrates as enactments migrate; Michigan SB 1050’s fate is the single best forward indicator of whether Standard moves toward the Restrictive column (passage pulls the modal template within two sessions, 65–75%).

3. Strictness is not desirability. The code grades authorization price. It takes no position on whether any price is correct.

4. Obligations compound across layers. Authorization price accumulates across all applicable, non-preempted layers; within each regulatory domain the most restrictive operative term controls, and across domains the obligations compound. A state’s grade is a floor — the Local, Regional, and Federal Overlays must be scored per site before any state grade is used for a siting decision.

5. Three instrument categories, never mixed. Binding — enacted or approved instruments — score on the Binding ledger. Frontier — active proposals, marked 🔶 — score separately and never enter Binding Strictness. Design precedent — failed, expired, or historical language, labeled as such in the tables — illustrates a mechanism and receives no score on either ledger.

Applied together, the five rules make any two graded jurisdictions directly comparable — same unit, same ledgers, same categories.

§4 — State Grading Output Format

A state’s grade publishes as an Authorization Profile, never as a single number:

Two states with identical composite strictness can present opposite commercial realities — a demanding but predictable Virginia (high certainty, high cost) against a permissive but locally volatile Indiana (low certainty, high friction) — which is exactly why the profile reports the dimensions separately.

Appendix A — Crosswalk to the Three-Cost Architecture

The twelve families roll up to the three costs every data center creates, preserving the reader-facing frame of the fifty-state baseline:

Appendix B — Provenance, Confidence, and Verification

The taxonomy consolidates two prior MindCast instruments: the three-cost, flexible-to-strict provision ladder from The Data Center Authorization Price (fifty-state baseline, T₀ = July 30, 2026) and the twelve-family canonical codebook with the 25 × (30/40/30) formula from the Washington strategy work. The merger preserves independent bargaining levers rather than collapsing them into a composite.

State anchors rest on enacted statutes, approved tariffs, effective executive orders, and binding local instruments verified through late July 2026; 🔶 marks pending, single-source, or aging placements requiring re-verification before external use. Confidence in tier placements: high (90%+) for jurisdictions with verified 2025–26 instruments; moderate (70–85%) for placements resting on pending bills. Session actions move weekly — re-verify any row before operational use, and log every instrument change to the ledger so tier migration is observed rather than inferred.

END OF ARTIFACT Model Data Center Authorization Code · Taxonomy v1.1 · MindCast AI, LLC · T₀ = July 30, 2026

See MindCast AI Data Center Regulation Live-Fire project

MindCast AI submitted a formal public comment on the Department of Energy’s draft 2026 National Transmission Needs Study. A docketed federal comment is an institutional intervention with a timestamp, and it doubles as a falsifiable test of our simulation work: we ask an agency for a specific action, and the agency’s response settles the prediction in the public record. Prior MindCast federal filings and government policy commentary are listed in the Appendix; the rest of this post is about the DOE issue, because it deserves the full space.

Our ask fits in one sentence: classify large AI loads by the net burden they create during the grid’s constrained hours, and credit flexibility only when the commitment behind it is enforceable, measurable, and financially backed. The sections below cover, in order, why an unglamorous triennial study carries this much leverage, what we asked DOE to change, what each possible DOE response would mean, and what the shift means for anyone building, financing, or regulating AI data centers.

Why a Draft Study Matters More Than Most Legislation

Congress requires DOE to publish the Needs Study every three years. The Study does not build anything, fund anything, or order anyone to do anything. The Study’s power runs through citation: regional transmission planners, state commissions, interconnection processes, and national corridor designations all reference it as the federal baseline for where the grid is constrained and where constraint is coming.

Whatever the final Study says about data center load becomes the default assumption in dozens of downstream proceedings through 2029. A definition adopted now compounds quietly for three years. Arguing with the baseline later costs far more than shaping it today — which makes the comment window, open until September 7, the cheapest leverage available in the entire AI infrastructure debate.

The draft also arrives carrying a finding that deserves far more attention than it has received. DOE estimates that the majority of national transmission congestion costs — roughly eleven billion dollars in 2023 — concentrate in about five percent of hours. The American grid’s binding constraint lives in a few hundred hours a year. Every argument about data centers and the grid changes shape once that fact sinks in.

The section reduces to one sentence: the Study sets the baseline, the baseline compounds for three years, and the congestion finding tells us exactly what the baseline should be measuring.

What We Asked DOE to Change

Our comment makes one central proposition: transmission planning should classify large AI loads by the net burden they create during system-constrained hours, and DOE should credit flexibility against identified need only when the commitment behind it is technically deliverable, contractually enforceable, continuously measurable, and financially backed.

Two facilities with identical nameplate ratings can impose materially different burdens on the grid, depending on whether they draw full power through a winter-storm peak or ride through it on batteries, on-site generation, and deferred workloads. Today’s needs methodology cannot see the difference. It measures load by interconnection-application size, which means it can overstate need where enforceable flexibility goes unrecognized and gives no planning value to the flexibility that could relieve constraints at lower cost than new transmission.

The second half of the proposition is the guardrail. Crediting flexibility in a needs assessment removes transmission from the national picture — infrastructure that will not be identified and eventually not built. A flexibility promise that fails during the exact hours it was credited against cannot be replaced in real time. We therefore proposed a seven-dimension commitment-quality screen — dispatchability, telemetry, duration, rebound management, testing, penalties, and financial assurance — and a performance-factor rule so that even enforceable commitments earn credit at their verified performance level, not at face value. Marketing-material flexibility earns zero.

Both halves together give DOE an instrument equal to its subject: need measured by how load actually behaves in constrained hours, and credit earned only by proof.

What a DOE Decision Would Mean

Three outcomes are possible, and each one moves the market.

If DOE adopts the framework in substance, large-load flexibility gains explicit and measurable value inside the federal transmission-needs framework. The Study would distinguish firm load from enforceably flexible load, model need under both, and report congestion-hour concentration as a standing metric. Every regional planner and state commission citing the Study inherits the distinction. Probability DOE adopts meaningful elements of net-burden classification or flexibility scenarios in the final Study or the next cycle: 35–50% — agencies rarely restructure a methodology in one comment round, though our filing record shows a precisely targeted request at the right procedural moment can move faster than the base rate suggests (Appendix).

If DOE adopts nothing, the comment still changes the record. The framework, the enforceability standard, and three falsifiable predictions now sit in a federal docket with a government timestamp. When the predictions resolve — and they resolve against public filings by July 2028 — the record shows who saw the structure early. Federal comment dockets provide one of the strongest public evidence-custody mechanisms available: independently timestamped, third-party held, permanently retrievable.

The intermediate outcome is the most likely and the most interesting. DOE acknowledges the firm-versus-flexible distinction qualitatively, and the harder crediting fight migrates to the venues already living it — state commissions and RTO tariff proceedings. Ohio’s data center tariff, Texas Senate Bill 6, and Virginia’s large-load rate class have already made enforceable financial and operational commitments a condition of service. A federal needs baseline that speaks the same language would harmonize the system; one that ignores it will measure a map the market has already redrawn.

Across all three outcomes, the filing pays. Adoption changes federal planning, silence preserves a timestamped record that the predictions will grade, and the intermediate path moves the crediting fight to venues where the framework already speaks the local language.

What This Means for AI Data Centers

For operators, investors, and regulators, the consequences reduce to three claims, each stated plainly below: workload mix is turning into money, enforceability is turning into currency, and unenforceable pledges are turning into nothing.

Workload architecture is becoming a financial variable. Training and deferrable batch compute can vacate constrained hours; latency-sensitive inference cannot. Once planning and tariffs price constrained-hour behavior, the workload mix inside a facility determines its interconnection speed, its tariff exposure, and its siting options. Facility design and portfolio strategy converge with grid strategy.

Enforceability is becoming the currency. Every layer of government is converging on the same trade: credible, penalty-backed, financially assured commitments in exchange for authorization — faster interconnection from utilities, service from state commissions, consent from communities. Operators who invest early in dispatch rights, telemetry, tested duration, and real financial assurance are building what we have called governance capability as competitive capital: the capacity to make commitments counterparties can verify, which is now worth megawatts and months.

Soft pledges are depreciating fast. “Grid-friendly” claims without enforcement mechanisms are already discounted by commissions and communities; a federal crediting standard would take their planning value to zero. The gap between facilities that can prove constrained-hour performance and facilities that can only assert it will show up in queue position, tariff terms, and ultimately cost of capital.

The common thread across all three claims: verified, enforceable constrained-hour behavior is becoming the unit of competitive advantage in AI infrastructure, and the market is pricing it before the federal government measures it.

The Predictions We Put in the Record

We staked three falsifiable predictions inside the comment, each resolving against public filings by July 2028. First: a majority of publicly disclosed or commission-filed hyperscale interconnection agreements above 100 MW executed after January 2027 will carry constrained-hour flexibility or on-site supply commitments (75–85%). Second: additional states and commissions will adopt standardized cost-causation rules for data centers (80–88%). Third: at least one RTO or commission will approve a tariff, pilot, order, or interconnection process expressly conditioning accelerated processing on verified, penalty-backed flexibility (70–80%).

Resolution cuts both ways, and we accept both edges. If the predictions resolve as stated, the load driving national need will have demonstrated in the public record the behavior we asked DOE to measure. If they miss, we publish the miss at the same prominence — the same discipline that governs every filing in this series.

The comment period closes September 7. Operators, utilities, and states each have a rare, cheap window to shape the federal baseline they will live under through 2029. We used ours.

MindCast AI produces institutional and behavioral analysis of AI infrastructure governance. The full DOE comment, including the seven-dimension commitment-quality framework and the complete prediction register, accompanies this post.

Appendix: Prior MindCast Government Filings and Policy Commentary

Federal Docket Filings

• U.S. Department of Justice — Live Nation remedies, Docket ATR-2025-0002-0023 (2025)

• Office of Science and Technology Policy — Regulatory Reform on Artificial Intelligence, Docket OSTP-TECH-2025-0067 (October 2025)

• U.S. Department of Justice and Federal Trade Commission — Updated Guidance on Collaborations Among Competitors, Docket ATR-2026-0001 (February 2026)

• Commodity Futures Trading Commission — Prediction Markets ANPRM, RIN 3038-AF65 (April 2026). The filing requested conversion to a Notice of Proposed Rulemaking within ninety days; the Commission delivered it in forty-one, carrying forward the recreation element of the gaming definition the filing proposed.

• Commodity Futures Trading Commission — Prediction Markets NPRM, RIN 3038-AF65 (June 2026), filed the day the proposal published.

Other MindCast Government Policy Work

• USPTO — Inter Partes Review Governance and Innovation (December 2025): how discretionary IPR denials operate as an innovation tax, with falsifiable predictions on assertion filings, settlement values, and AI-stack enclosure risk through Q4 2026.

• Federal enforcement and the states — Federal Political Market Failure and State Substitution as a Free-Market Corrective (January 2026): why state attorneys general function as competing suppliers of enforcement when federal institutions fail, with a Critical De-Risking Zone framework for post-clearance merger exposure.

  ---

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

MindCast AI Infrastructure Live-Fire

Executive Summary

America is no longer debating whether AI infrastructure should be built. It is competing over which governance architecture will authorize it.

The evidence arrived in Congress this year. Senator Bernie Sanders introduced the Artificial Intelligence Data Center Moratorium Act in March 2026; Representative Alexandria Ocasio-Cortez introduced the House companion in June. Together, S.4214 and H.R.9442 — the most aggressive AI infrastructure bill in American history — would halt construction of every large AI data center in the country until Congress passes sweeping safeguards covering jobs, electricity prices, water, privacy, and community consent. Most coverage frames the bills as a binary choice: pause AI infrastructure or accelerate it. The framing misses the real event.

Three governance systems now compete for control of that question, and each is winning somewhere. The Moratoriumequilibrium — pause deployment until safeguards exist — dominates county boards and one state capital, but cannot pass Congress. Pure Acceleration — permit quickly with minimal conditions — dominates federal executive policy, but keeps colliding with state and local resistance. Conditional Acceleration — keep building, but make developers pay the public costs they create — is emerging as the system’s dominant coordination equilibrium, with both parties now writing versions of it into law.

Two findings drive the analysis. First, conditional acceleration will prevail — not because it polls better, but because the system’s own feedback loops push every conflict toward it. Second, and more surprising: the moratorium bill’s most powerful effect is strengthening the equilibrium it opposes. A credible pause threat gives developers a reason to negotiate, utilities a reason to build dedicated rate structures, and communities leverage to demand enforceable terms — all of which are conditional acceleration in action.

MindCast tested the thesis by running its proprietary Cognitive Digital Twin Foresight Simulation engine — computerizing behavioral economics and dynamic predictive game theory — across the full institutional field. The simulation confirmed both findings and sharpened the first: conditional acceleration is not a compromise between the poles but a coordination equilibrium — the arrangement that lets institutions with incompatible objectives keep authorizing infrastructure through priced, divisible obligations instead of ideological agreement. Headline results:

• ⚖️ Conditional acceleration becomes America’s dominant practical governance architecture through July 2028 (85–92%)

• 🔌 Federal acceleration channels projects into state and local bargaining rather than eliminating it (86–93%)

• 📜 Large-load tariffs and cost-allocation frameworks standardize across major AI markets (82–90%)

• 🏗️ Governance capability becomes a competitive differentiator — by 2028, at least two hyperscale developers compete publicly on governance architecture itself (80–89%)

Twelve falsifiable predictions — five primary, seven secondary — form the paper’s register, each with a deadline, a confidence band, and a public condition that proves it wrong. The analysis is written for the people who will operate inside this system — hyperscale developers, state and federal lawmakers, and data center strategists — and it does more than forecast: Section IX translates the findings into moves for each seat at the table, and Section XII redesigns the moratorium bills under the paper’s own framework, a legislative blueprint for the conditional-acceleration center.

One claim differentiates the analysis from conventional policy debate, and it frames everything below: governance equilibria compete through institutional adaptation, not legislative victory. Bills matter for how they shift every other institution’s incentives, not for whether they pass. A second contribution follows from the first: federal and state proposals are complements, not substitutes — Congress contests the quantity of AI infrastructure while states set its price — and conditional acceleration is the architecture that links those nested layers into one functioning system.

The Three Equilibria at a Glance

One table organizes the entire contest. Three governance architectures now compete for AI infrastructure, and each reduces to a decision rule, a mechanism, and a trajectory:

Carry the map forward: each row is a complete governance architecture, already operating at the level of government where it is strongest.

What This Paper Is — and Is Not

Written for readers on every side of the moratorium debate, the paper advocates for no bill and against no bill. The analysis does not argue that AI infrastructure should proceed without governance, and it does not argue that nationwide moratoria represent the optimal governance architecture. The task is different: evaluating competing institutional mechanisms for reconciling AI competitiveness with the increasingly visible public costs of hyperscale infrastructure. The central question is predictive rather than ideological — which governance architecture is most likely to sustain both innovation and political legitimacy over time?

The forecasts are built to be falsifiable, and the falsifiability is the invitation. Readers who believe a different architecture will prove more stable — a stronger federal backstop, a faster federal track — can test that belief against the prediction register, where every entry carries a deadline, a confidence band, and a public condition that proves it wrong. Engagement on mechanisms and forecasts, rather than on positions, is what the paper asks of every reader.

I. Governance Has Become the Competitive Arena

America’s AI debate changed subject in 2026, and the change happened faster than most observers registered. Regulators stopped asking software questions — what do the models do, who do they harm — and started asking infrastructure questions: how much power does the facility draw, whose water does it use, whose electric bill rises.

Earlier MindCast analyses tracked the shift in two stages. Governments began regulating AI’s physical production — electricity, transmission, water, land use, environmental review — with the legal machinery built for power plants, not software. The political negotiation then narrowed to a single question: not whether AI gets built, but who bears the costs of building it. New York’s statewide permitting pause, Seattle’s data center freeze, and more than one hundred county moratoria all answer that question locally, each on its own terms.

The Moratorium Act reveals a third stage. Competing institutions no longer fight over individual policies; they fight to establish entire governance architectures — complete systems of rules, prices, and veto rights that, once established, reproduce themselves. Understanding those architectures, rather than handicapping any single bill, is the work of the rest of the paper.

II. Institutions Optimize Different Problems

Conflict over AI infrastructure looks chaotic until each institution is viewed through its own optimization problem. Viewed that way, the chaos resolves into five rational actors solving five different problems on the same board.

The Federal Executive optimizes national competitiveness: every month of permitting delay reads as strategic ground ceded to China. Congressional moratorium advocates optimize precaution and democratic oversight: irreversible harms, in their model, justify pausing before building. Hyperscale developers optimize speed and certainty: delay costs more than concessions. Utilities optimize reliability and cost recovery: growth must not destabilize existing customers or shift costs onto them. State and local governments optimize both sides at once: capture the investment, protect the residents, survive the next election.

None of these objectives is irrational. They are simply different — and because each institution holds real power over some part of the approval pathway, each pushes the whole system toward a different governance equilibrium. The architecture that ultimately governs will therefore emerge from continuous adaptation among these institutions, not from any single legislative outcome. Section III names the three equilibria those adaptive pressures produce.

III. Three Governance Equilibria

Three complete governance architectures now compete for American AI infrastructure, and each can be stated as a one-line decision rule. The table at the opening of the paper summarizes them; the paragraphs below make each one concrete.

The Moratorium equilibrium pauses first and asks questions later. The Sanders–Ocasio-Cortez bills carry it federally, New York’s Executive Order 62 carries it at the state level, and a wave of county and city freezes carries it beneath them. Grievance politics fuels the equilibrium: residents weigh feared losses — higher bills, drained aquifers — roughly twice as heavily as promised gains, and a pause is the only instrument that removes every feared loss at once. Local strength and national weakness share the same source: a county board answers directly to those residents, while a federal moratorium must overcome labor opposition, the China-competitiveness argument, and the constitutional reality that land use belongs to states.

The Conditional Acceleration equilibrium preserves deployment by converting uncompensated externalities into negotiated obligations. Every instrument in its toolkit manifests that single principle: cost-causation tariffs convert grid impacts into payments, dedicated large-load rate classes convert ratepayer exposure into insulation, community-benefit agreements convert local harms into escrowed and independently monitored commitments, and financial assurances convert stranded-infrastructure risk into posted guarantees. The coalition crosses party lines: New York’s Democratic governor demands a grid fund, a Republican congressman from data-center country demands full cost payment and water recycling, and twenty-seven state legislatures are drafting between them.

The Pure Acceleration equilibrium clears the road. Federal instruments power it: a 2025 executive order that streamlines environmental review, opens federal land, and fast-tracks projects above 100 megawatts, plus federal energy regulators pressing all six regional grid operators to speed large-load connections. The limit is structural rather than political: federal orders cannot erase state zoning or county land-use power, so every project acceleration delivers another facility to a local government that acceleration cannot compel.

Read as a set, the three equilibria are not competing bills awaiting a vote. They are adaptive governance systems, each already operating at the level of government where it is strongest — and Section IV shows what happens when they interact.

IV. Adaptive Competition

Governance regimes rarely defeat one another outright. They adapt — and in AI infrastructure, every institutional decision changes the incentive landscape facing every other participant.

Concrete examples run in every direction. Federal permitting acceleration pushes more projects into state review at once, which drives states to negotiate conditions rather than issue blanket prohibitions. State bargaining pushes developers to finance transmission, private generation, and community investments before opposition organizes. Utility cost allocation redraws developer site-selection maps. Community resistance reshapes utility tariffs. Congressional moratorium proposals redraw the political boundaries inside which everyone else negotiates.

Two feedback loops carry the whole system, and both deserve explicit statement. The concession loop runs: infrastructure demand raises grid stress; grid stress forces utilities to allocate costs; visible costs fuel local resistance; resistance extracts developer concessions; concessions buy political acceptance; acceptance restores deployment. The channeling loop runs: federal acceleration raises deployment velocity; velocity intensifies state bargaining; bargaining spreads conditional-acceleration frameworks — which simultaneously drain political energy from a nationwide moratorium and constrain pure acceleration from below.

Both loops end in the same place, and the convergence is the paper’s structural core. Conditional acceleration is not merely the most likely outcome; it is the system’s attractor — the state both feedback loops push toward regardless of where a conflict starts. A moratorium fight feeds the concession loop and produces negotiated terms. An acceleration push feeds the channeling loop and produces state conditions. Either way, the system lands on priced, negotiated deployment.

No equilibrium will win nationally, because each controls a different layer of government — a conclusion that deserves plain statement. Acceleration governs federal permitting, conditioning governs state and utility policy, and pauses govern a growing share of counties — and for any given project, the strictest rule that no higher authority has overridden controls the outcome. The national result is a composite, not a victory. Section V maps the layers formally, because the map resolves what otherwise looks like national incoherence.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

V. The Governance Stack: Why Federal and State Proposals Play Different Games

Federal and state proposals look contradictory at first read — Congress debates pausing the very industry twenty-seven state legislatures are busy pricing. Read as rivals for the same authority, the record is incoherent. Read as layers of a governance stack, it snaps into focus: federal and state proposals are adaptive responses at different levels of the system, solving different optimization problems.

Five layers structure the stack, and each asks its own question:

Every layer governs a different margin. Federal institutions primarily determine the national pace of deployment. States determine the economic conditions under which deployment occurs. Utilities determine cost allocation. Local governments determine project legitimacy. Developers determine whether the accumulated conditions remain economically acceptable.

Reduced to economics, the division is quantity versus price. Congress contests how much AI infrastructure America wants — the moratorium bills argue for less now, federal acceleration argues for more, faster. States set the authorization priceof whatever quantity arrives: grid funds, large-load tariffs, water obligations, community consent. Different questions explain otherwise puzzling behavior — the same hyperscaler opposes the federal pause while testifying in state hearings that it will pay full infrastructure costs, without contradiction, because the two proposals tax different margins.

Congress determines how much AI infrastructure America wants. States determine its authorization price.

The layers do not merely coexist; they continuously rewrite one another’s bargaining leverage. A live federal pause debate raises the expected cost of resisting state terms, so developers concede more at state tables — federal threat becomes state leverage. Enacted state frameworks then let Congress point to functioning governance, draining urgency from the national moratorium — state success becomes federal deflation. Failed state bargaining — a rate spike, a water controversy — refuels it. The full chain runs recursively: federal proposals change state incentives, states change developer behavior, developer behavior changes federal politics, and federal politics changes the next round of state legislation. Governance evolves recursively, not hierarchically.

The recursion produces a structure the three-equilibrium framework alone cannot name: a Nested Governance Equilibrium. Within it, the federal government pursues acceleration as industrial policy; states specialize in allocating infrastructure costs and defining statewide conditions; utilities operationalize those conditions through tariffs, interconnection rules, and cost recovery; local governments negotiate place-specific legitimacy through zoning and community-benefit agreements; and developers optimize across all four public layers by treating governance capability as core competitive capital.

Conditional acceleration then becomes more than one equilibrium among three. It becomes the coordinating architecture that links the nested layers into a functioning system — the explanation for why different levels of government can pursue apparently conflicting policies while collectively producing a stable national outcome. The moratorium bills’ role inside that recursion is the subject of the next section.

VI. The Moratorium’s Strategic Importance

The Moratorium Act will almost certainly never govern American AI infrastructure — and it still matters more than most bills that pass. Understanding why requires separating the legislation’s text from its function.

The text is brief and maximal. The bills define a covered facility by physical signature — twenty megawatts of draw, twenty-kilowatt racks, liquid cooling, with commonly owned sites counted together — and freeze construction and upgrades on enactment.

The freeze lifts only when Congress passes a safeguard package spanning federal pre-market review of AI products, job-loss prevention, wealth sharing, guaranteed non-increase of consumer bills, binding community approval, subsidy prohibition, and union labor. No Congress can plausibly deliver that package as one act, so the pause would run indefinitely. Nine House Democrats cosponsored at introduction; neither chamber has scheduled a vote, and none is likely.

The function is bargaining power, distributed to everyone except the sponsors. A credible pause threat gives developers a stronger incentive to negotiate terms rather than resist them. Utilities gain cover to build dedicated rate structures. States gain political support for requiring infrastructure contributions. Communities gain leverage to demand enforceable commitments instead of press-release promises. Every one of those effects strengthens conditional acceleration — the equilibrium the bill’s sponsors reject.

The bill’s own text concedes the endgame. Buried in its exit conditions sits the requirement that new facilities prove they raise no consumer’s electric bill — a cost-internalization rule wearing moratorium clothing. Both parties have already converged on that principle; they disagree only about pausing while it gets written.

Two provisions will likely outlive the vehicle entirely. Export controls on AI hardware to countries without comparable safeguards serve a national-security agenda far larger than the pause coalition, and the bill’s quarterly facility-transparency reports — water, energy, emissions, noise, subsidies — match the monitoring architecture that durable local bargains require. Expect both to resurface in other bills under other sponsors, because pressure movements that cannot win their headline demand reliably migrate to their strongest severable ones.

VII. Why Conditional Acceleration Occupies the Institutional Center

Neither pole can hold the country, and the reasons are symmetrical. Pure acceleration keeps colliding with local resistance over electricity prices, transmission capacity, water, and financing — resistance no federal order can override. Comprehensive moratoria collide with equally hard limits: delayed investment, lost jobs, and a national-competitiveness argument that peels away every marginal vote.

Converting externalities into obligations is why conditional acceleration absorbs pressure from both directions: the conversion gives every major participant most of what it wants. Projects proceed, which satisfies developers and the federal executive. Communities receive enforceable compensation, which defuses the grievances that fuel moratoria. Utilities recover costs through dedicated rate classes, which protects existing customers. Developers trade higher costs for something more valuable: permitting certainty and durable local acceptance.

The equilibrium also reframes the industry’s real challenge. The question is no longer whether AI infrastructure should exist. The question is whether developers can construct enough political legitimacy — through payment, protection, and enforceable commitment — to sustain deployment for a decade. Conditional acceleration is simply the name for governance that makes legitimacy purchasable on known terms.

VIII. From Static Regulation to Adaptive Governance

Behind the specific contest sits a conceptual shift that outlasts it. Traditional regulatory analysis assumes a sequence: governments set rules first, markets respond second. AI infrastructure demonstrates the reverse — markets, utilities, governments, and communities adapt to one another continuously, and no rule stays settled long enough to be a fixed starting point.

Evidence accumulates weekly. Every permitting decision changes the next negotiation’s baseline. Every tariff filing influences the next quarter’s site selection. Every community agreement resets expectations in the next county. Governance has become an evolving strategic process rather than a fixed framework, and the institutions that thrive are those that adapt fastest without losing their identity.

Recursive adaptation, not centralized design, is now the operating system of American AI infrastructure policy — which is why the next section forecasts emergence rather than enactment.

IX. Implications for the Next Decade

No single congressional vote will decide the future of American AI infrastructure. The outcome will emerge from thousands of interconnected negotiations running simultaneously across federal agencies, state governments, utilities, county commissions, developers, and local communities.

Each institution will keep optimizing its own objectives, and each adaptation will reshape the incentives facing every other participant. Emergent order of that kind favors the architecture that asks the least total sacrifice — and conditional acceleration holds that position, because it partially satisfies every major participant without requiring any of them to achieve complete victory.

The implications land differently in each seat at the table — but two syntheses hold across all of them.

Policymakers at every level: governance durability depends less on choosing between acceleration and restriction than on designing institutions that allocate infrastructure costs transparently, protect affected communities, and preserve long-term investment incentives.

Developers of every scale: long-term deployment depends as much on authorization capability as on engineering capability — firms that secure durable authorization through credible commitments, utility coordination, and community engagement will deploy more predictably than firms relying on capital or federal acceleration alone.

The seat-specific moves follow.

Hyperscalers and developers: governance capability now compounds like compute — build the bargaining function, deliver enforceable terms before opposition organizes, and treat durable authorization as an asset to acquire.

State lawmakers: conditional acceleration is yours to standardize, and the state that writes the model large-load framework first sets the template competitors copy.

Federal lawmakers: neither the pause nor preemption can pass, but cost-allocation standards and transmission reform can — legislate where the bipartisan center already exists, and Section XII shows how to redesign the moratorium bills into that center.

Data center strategists and investors: price authorization risk jurisdiction by jurisdiction, because the strictest surviving local rule, not the federal posture, decides each project.

The practical forecast follows directly. Expect negotiated conditions to spread faster than either pauses or unconditioned fast-tracks; expect the price of a megawatt to include, increasingly, the price of local legitimacy; and expect the moratorium movement’s real legacy to be the enforceable terms its threat extracted. Section X reports what MindCast’s simulation found, and the prediction register puts each expectation on a public clock.

X. What the Simulation Found

MindCast tested the three-equilibrium thesis by running its proprietary Cognitive Digital Twin Foresight Simulation engine — computerizing behavioral economics and dynamic predictive game theory — across the full institutional field: the federal executive, the congressional moratorium coalition, hyperscale developers, utilities and grid operators, and state and local governments, each modeled as an adaptive behavioral system rather than a stakeholder holding positions. Six findings emerged, and each sharpens the paper’s argument.

Conditional acceleration is a coordination equilibrium, not a compromise. The simulation’s central result reframes the middle path. Compromises require participants to abandon objectives; coordination equilibria let participants keep incompatible objectives and transact anyway. Conditional acceleration holds the strongest coordination integrity in the field because it runs on divisible transactions — a tariff here, a water guarantee there, a community fund at the next site — rather than on ideological agreement anyone must sign.

The governing game shifts in sequence. The simulation shows the contest moving from Build-versus-Block to Price-and-Condition, and ultimately toward standardized public bargains — template frameworks that make the price of authorization knowable before a project is announced.

Authority fragments; it does not concentrate. No single forum — Congress, the federal executive, energy regulators, statehouses, county boards — captures control of the authorization pathway. Governance capability, not forum capture, therefore decides outcomes, because the winning institution must transact across every forum at once.

Authorization throughput, not compute, constrains innovation. Durable national AI advantage depends on how fast the system can convert applications into authorized, accepted capacity. A country that accumulates chips faster than permissions stalls exactly where the United States is stalling now.

Every actor holds adaptive fallbacks, not fixed objectives. The moratorium coalition falls back to disclosure, worker protections, and community compensation when the pause loses viability. Developers fall back to private generation and geographic diversification when costs rise. The fallback ladders — not the headline positions — determine where the system settles.

Adaptive coherence favors hyperscalers and utilities. Evaluated for which institutions preserve coherent decision-making as the game changes, hyperscalers and utilities score strongest, because both convert changing constraints into repeatable bargaining mechanisms rather than fighting each constraint anew. Governance capability functions as productive capital — and the competitive frontier shifts from acquiring compute to acquiring durable authorization.

The last finding deserves more than a paragraph, because it moves the paper beyond policy analysis into innovation economics. Section XI elevates it, and the prediction register puts every finding on a public clock.

XI. Governance Capability Becomes Competitive Capital

One simulation finding deserves promotion from result to thesis: governance capability is becoming productive capital — an asset that compounds, differentiates competitors, and earns returns, exactly as compute did before it.

Under pure acceleration, competitive advantage derives from capital, technology, and deployment speed. Under conditional acceleration, another asset becomes scarce: the ability to obtain durable authorization. Developers no longer compete only on capital, land, electricity, and GPUs. They increasingly compete on authorization, community bargaining, utility negotiation, infrastructure finance, and governance architecture itself.

The conversion mechanism is concrete. Developers capable of negotiating durable community agreements, financing infrastructure credibly, allocating public costs transparently, and repeating authorizations across jurisdictions convert governance itself into productive capital. Authorization stops functioning as a compliance obligation and starts functioning as a competitive capability — one that prices like a portfolio asset, because a performed agreement in one county lowers the discount the next county applies to every promise, while a breach raises it everywhere at once.

AI infrastructure therefore begins rewarding institutional competence alongside engineering competence. The developer who can repeat a credible public bargain holds an advantage no GPU allocation can replicate — and unlike chips, governance capital cannot be bought on a spot market; it must be built, performed, and defended deal by deal.

Prediction TGE-4 puts the claim on a clock: by 2028, expect at least two hyperscale developers to compete publicly on governance architecture itself.

XII. Improving the Moratorium Framework

The Artificial Intelligence Data Center Moratorium Act correctly identifies an emerging governance problem, and the diagnosis deserves to be separated from the mechanism. Hyperscale AI infrastructure imposes increasingly visible demands on electricity systems, transmission networks, water resources, and surrounding communities — externalities of that scale require governance, not neglect. On the diagnosis, the sponsors are right. The legislation, however, adopts prohibition as its coordination mechanism, and the simulations show why prohibition cannot become the dominant national equilibrium: it attempts to resolve a bargaining problem through suspension rather than structured negotiation.

Three design principles convert the bills into durable governance.

Principle one: replace prohibition with authorization conditions. Instead of pausing new AI data centers, Congress could let projects proceed upon satisfying objective criteria: demonstrated cost causation for new transmission and distribution infrastructure, protection against residential ratepayer subsidization, water-management and recycling plans suited to local conditions, financial assurance for long-lead investments, transparent reporting of electricity, water, and community commitments, and enforceable community-benefit agreements where appropriate. Authorization conditions deliver every public protection the moratorium promises — without the pause.

Principle two: standardize governance instead of negotiating every project from scratch. Project-by-project bargaining raises developer uncertainty while consuming institutional capacity that states and counties do not have. A federal framework of predefined performance standards — with states and utilities retaining implementation flexibility, and compliant projects earning greater permitting certainty — lowers transaction costs without eliminating state and local authority, the constitutional lane the courts have left open.

Principle three: maximize authorized infrastructure, not infrastructure. The legislative objective should be neither to maximize nor to minimize AI infrastructure but to maximize authorized infrastructure — capacity that proceeds with durable political legitimacy, transparent cost allocation, and repeatable governance mechanisms. Infrastructure authorized on those terms outvalues both infrastructure delayed indefinitely and infrastructure accelerated without sustainable consent.

The redesign changes what the paper offers its readers. Sponsors keep their diagnosis and gain a mechanism that can actually pass; developers, utilities, congressional staff, and governors who reject the equilibrium forecast still receive concrete design principles grounded in the same framework. The register below then puts the framework’s expectations — redesigned or not — on a public clock.

Prediction Register — TGE

MindCast publishes forecasts the way an exchange lists contracts: each entry carries a deadline, a confidence band, and a condition that proves it wrong. Every entry below resolves from public records — statutes, utility-commission dockets, municipal records, corporate disclosures, and the Congressional Record — and the register stays live through July 2028, with resolutions published as they land.

Primary Predictions

TGE-1. Conditional acceleration becomes the dominant practical U.S. governance architecture for AI infrastructure. Deadline: July 2028. Confidence: 85–92%. Falsifier: statewide moratoria pass in five or more states, or states enacting large-load cost frameworks fail to outnumber states enacting statewide moratoria by at least three to one.

TGE-2. Large-load tariff and cost-allocation frameworks standardize across major AI markets. Deadline: July 2028. Confidence: 82–90%. Falsifier: large-load terms remain predominantly bespoke, project by project, across the Virginia, Texas, Georgia, and Ohio corridors.

TGE-3. Moratorium policy payloads — hardware export controls, facility transparency mandates, worker and community protections — migrate into narrower legislative and regulatory vehicles. Deadline: July 2028. Confidence: 76–86%.Falsifier: no payload appears in a vehicle outside the moratorium coalition’s sponsorship.

TGE-4. Governance capability becomes a competitive differentiator: by 2028, at least two hyperscale developers publicly compete on governance architecture — standardized community-benefit agreements, transmission funding, water commitments, or similar authorization frameworks — rather than relying solely on technical or financial advantages. Deadline: July 2028. Confidence: 80–89%. Falsifier: fewer than two hyperscale developers publicly deploy standardized governance frameworks as competitive differentiators, and community strategy remains housed in public relations across the top operators.

TGE-5. Federal acceleration channels projects into downstream state and local bargaining rather than eliminating it. Deadline: July 2028. Confidence: 86–93%. Falsifier: Congress enacts a federal siting preemption statute, or federally fast-tracked projects bypass state and local terms at scale.

Secondary Predictions

TGE-6. Utilities become the primary operational regulators of AI infrastructure, with rate cases and interconnection terms deciding more outcomes than zoning fights. Deadline: July 2028. Confidence: 72–84%. Falsifier: contested sitings resolve primarily in land-use and environmental forums, with utility terms incidental to outcomes.

TGE-7. Local moratoria increasingly function as bargaining mechanisms rather than permanent prohibitions — enacted, negotiated against, and lifted on terms. Deadline: July 2028. Confidence: 70–82%. Falsifier: a majority of local moratoria harden into permanent prohibitions rather than expiring into negotiated frameworks.

TGE-8. Political language shifts from “whether to build” toward “under what conditions projects are authorized.” Deadline: July 2028. Confidence: 68–80%. Falsifier: hearing records, party platforms, and campaign materials in data-center jurisdictions remain dominated by build-versus-block framing.

TGE-9. Republican and Democratic jurisdictions converge on similar cost-allocation mechanisms despite different rhetoric. Deadline: July 2028. Confidence: 72–84%. Falsifier: enacted cost-allocation mechanisms cluster overwhelmingly in one party’s jurisdictions.

TGE-10. Fragmented bargaining drives demand for standardized governance templates — model frameworks, uniform tariff structures, multi-state coordination. Deadline: July 2028. Confidence: 74–85%. Falsifier: no model frameworks or standardization efforts emerge, and terms remain fully bespoke as volume grows.

TGE-11. A major reliability, ratepayer, or water controversy temporarily strengthens moratorium rhetoric without producing a lasting national pause. Deadline: July 2028. Confidence: 65–78%. Falsifier: such a controversy produces an enacted, durable national or multi-state permanent pause.

TGE-12. Authorization markets emerge: developers begin competing for jurisdictions on authorization throughput rather than electricity prices and tax incentives alone, while states begin competing by advertising faster authorization, standardized governance, repeatable permitting, and predictable community-benefit frameworks. Deadline: July 2028. Confidence: 74–86%. Falsifier: site-selection announcements and state recruitment materials continue to feature only power costs and tax incentives, with no authorization-throughput or governance-certainty marketing.

Scoring is public by design. Each resolution will publish as it lands — hits and misses at the same level — because a register that hides its misses is marketing rather than foresight.

Conclusion

Congress introduced a moratorium; history will record a bargaining position. The Artificial Intelligence Data Center Moratorium Act is best understood not as legislation likely to pass but as evidence that AI infrastructure governance has entered a new phase — one where institutions compete to establish entire governance architectures rather than individual rules.

Three architectures now run simultaneously. One prioritizes precaution through delay and holds the counties. One prioritizes speed through federal coordination and holds the executive branch. One aligns private incentives with public costs — and holds the center, the feedback loops, and the future. Linked through the governance stack, the three form a nested equilibrium: apparently conflicting policies at different layers of government, collectively producing one stable national outcome.

History supplies the precedent. Railroads, interstate pipelines, electric transmission, telephone networks, and broadband each began with unconditioned private buildouts, provoked public backlash over who paid, and settled into governance architectures that priced public externalities rather than permanently prohibiting investment. AI infrastructure is entering the same historical pattern — which makes conditional acceleration not a momentary political compromise but the latest instance of a recurring American governance trajectory.

The introduction of the Artificial Intelligence Data Center Moratorium Act demonstrates that the governance debate has entered a new phase. The central policy question is no longer whether AI infrastructure requires governance, but which governance architecture can most effectively align innovation, public legitimacy, and durable authorization. Conditional acceleration is this paper’s answer — the most likely, not the only conceivable — and the prediction register is a standing invitation: policymakers who believe another architecture will prove more stable can test that proposition against the same mechanisms and deadlines. For the companies racing to deploy, meanwhile, the frontier has already moved: the first generation of AI winners acquired compute, and the next will acquire durable authorization.

Appendix A — MindCast Proprietary CDT Foresight Simulation Register

MindCast evaluated the three-equilibrium contest through its proprietary Cognitive Digital Twin Foresight Simulation engine, computerizing behavioral economics and dynamic predictive game theory. Each institutional twin is specified on an eight-field schema — mental model, objectives, constraints, relationships, decision variables, observables, adaptation rules, and dynamics created — so that every actor functions as a behavioral model generating decisions, not a stakeholder holding positions.

Full state specifications remain internal; simulation outputs are analytical projections rather than observed outcomes. Section X reports the integrated findings; the register below records the twins.

CDT-1 — Federal Executive. Mental model: AI compute is strategic infrastructure; time lost is strategic disadvantage. Adaptation rules: opposition rises → negotiated mitigation rises; China narrative strengthens → acceleration rises; grid instability rises → infrastructure funding rises.

CDT-2 — Congressional Moratorium Coalition. Mental model: unchecked deployment creates irreversible externalities. Adaptation rule: when the moratorium loses viability, the demand ladder shifts to disclosure → worker protections → environmental safeguards → community compensation. The ladder generates prediction TGE-7.

CDT-3 — Hyperscaler. Mental model: delay is expensive; negotiation is cheaper than prohibition. Adaptation rules: opposition rises → voluntary commitments rise; power costs rise → dedicated generation; regulation fragments → geographic diversification.

CDT-4 — Utility and Grid Operator. Mental model: growth must not destabilize existing customers. Adaptation rules: demand accelerates → dedicated rate classes; backlash rises → cost allocation tightens.

CDT-5 — State and Local Government. Mental model: AI investment creates prosperity and political risk simultaneously. Adaptation rules: opposition rises → developer obligations rise; interstate competition rises → incentives rise.

System CDT — Governance Equilibrium Competition. The system twin simulates regimes rather than organizations: each regime carries its Section III decision rule plus a response set and an influence set, and regime viability is endogenous to the five institutional twins’ adaptive behavior. The two loops in Section IV form the register’s state-transition backbone.

Appendix B — External Sources Cited

External sources below ground every factual claim in the paper; none is required reading to follow the argument.

• Congress.gov, S.4214 (introduced Mar. 25, 2026) and H.R.9442 (introduced June 24, 2026), Artificial Intelligence Data Center Moratorium Act, 119th Congress.

• The White House, Executive Order 14318, “Accelerating Federal Permitting of Data Center Infrastructure” (July 2025).

• FERC, large-load show-cause orders to all six regional grid operators (June 2026).

• N.Y. Executive Order 62, statewide data-center permitting pause (July 14, 2026).

• Rep. Michael Baumgartner, Power and Water for Families Act (2026).

• Murphy v. NCAA, 584 U.S. 453 (2018) — limits on federal preemption of state law.

Appendix C — MindCast AI Works Cited

Prior MindCast analyses supply the frameworks this paper extends. Each stands alone; relevance to the present argument is stated for every entry.

Foundational

• New York’s Data Center Moratorium — How States, Cities, and Counties Now Regulate AI Data Center Development — Direct predecessor. Establishes the reclassification of AI as regulated infrastructure, the vertical governance stack and its “strictest non-preempted layer” rule, the incumbent-windfall mechanics of grandfathered pauses, the “charge them, don’t pause them” position, and the AIRC-III prediction register this paper’s TGE entries extend.

• The Two-Ledger Data Center Bargain — Why Federal Acceleration Makes Local Cost-Benefit Negotiation the Binding Constraint — Supplies the negotiation engine beneath conditional acceleration: loss-aversion weighting, the portable-grievance effect, the pre-coalition timing window, the Federal Acceleration Paradox, and the TLSM register. Section IV’s feedback loops formalize dynamics first surfaced in its simulation.

• Predictive Game Theory Meets the Era of AI — Operationalizing Fudenberg with Cognitive Digital Twins — The methodological foundation. Documents how MindCast computerizes behavioral economics and dynamic predictive game theory through Cognitive Digital Twins, including the game-replacement framework and the Adaptive Coherence Equilibrium concept that Section X applies to hyperscalers and utilities. See also its cited sub-links for the full method lineage.

Supporting

• The Power Stack — How Energy Infrastructure Became the New AI Battleground — Sequenced the resource hierarchy (compute → capacity → power → transmission) that this paper’s bargaining layer now terminates, and forecast the trigger conditions for government intervention that New York’s order satisfied.

• The AI Infrastructure Energy Opportunity Landscape — Supplies the deployment-path availability test — does a rule increase or decrease the number of places a project can viably be built — that separates conditional acceleration from the moratorium in Section VII.

• The Federal-State AI Infrastructure Collision — Forecast partial federalization: federal control of process, state control of place and cost. Grounds the finding that pure acceleration is federally strong but subnationally self-limiting.

• The Commerce Clause as America’s AI Advantage — Establishes the constitutional boundary on federal preemption — absorption, not erasure — behind prediction TGE-5 and the conclusion that no siting preemption statute passes.

• Synthesis in National Innovation Behavioral Economics and Strategic Behavioral Coordination — Quantifies institutional throughput and the temporal mismatch between market, government, and infrastructure cycles, grounding Section X’s finding that authorization throughput, not compute, constrains innovation.

• AI Repricing Cycle 2026 — Establishes the repricing-cycle framework in which authorization becomes the fourth scarce input, the market context for treating governance capability as productive capital.

• Chevron–Microsoft Project Kilby and the Firm-Power Forecast — Documents jurisdiction selection as a regulatory routing strategy, the precedent for the developer fallback behaviors (private generation, geographic diversification) modeled in CDT-3.

• Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure — Describes the runtime simulation infrastructure behind the engine that produced Section IX’s findings and the register’s confidence bands.

Registry namespace: TGE. Predictions carry explicit confidence bands and falsifiers, resolving against public records through July 2028.

Executive Summary

On Friday, July 24, thirty-five technology organizations — NVIDIA, Microsoft, Meta, OpenAI, ServiceNow, and Palantir among them — published an open letter urging Washington not to restrict open-weight AI models: models anyone can download, modify, and run on their own machines. Jensen Huang launched it with his first-ever post on X. Eleven million views later, the letter stands as the AI industry’s most visible policy statement of the year.

Public debate treats the letter as one side of a war between open and closed AI. Read the signature list instead of the prose, and a different story appears. Nearly every signatory sells something that becomes more valuable when models become cheap: NVIDIA sells the chips, Microsoft the cloud, ServiceNow the workflow software, CrowdStrike the security layer. The two most prominent absentees tell the same story from the other side: Anthropic’s revenue depends heavily on frontier capability staying scarce, and Google’s accelerators are rented through its own cloud rather than sold as merchant hardware — so open deployment benefits Google only indirectly, and NVIDIA far more cleanly.

The signature pattern is this paper’s subject, and it follows a law economists have understood since David Teece’s 1986 work on who profits from innovation. When a critical input becomes abundant, value does not disappear — it moves to whatever the input still needs in order to be useful. Open weights are making model capability abundant. The money is therefore moving to the complements: computing infrastructure, enterprise software, compliance and certification, proprietary data, and institutional trust. The letter describes the diffusion; this paper follows the money.

Six predictions anchor the analysis, each dated and scored so it can fail in public:

• Neither open nor closed models win. Enterprises settle into a lasting hybrid: frontier models for the hardest problems, open weights for everything else. 80–85% confidence, by end of 2028.

• The picks-and-shovels layer outgrows the model layer. Infrastructure, enterprise software, and compliance categories grow revenue faster than model makers do. 65–70%, by mid-2028 reporting.

• Trust and the right to operate hold their price longest. Certification, insurance qualification, and institutional relationships keep premium pricing for eight-plus years, while routing software and generic compliance tooling decay inside four. 75–80%.

• Washington regulates by nationality before it regulates by capability — restricting Chinese models before defining how capable any model must be to warrant restriction. 65–70%, by mid-2027.

• A currently unsigned frontier lab — Anthropic or Alphabet — buys or builds a picks-and-shovels business. 70–75%, by end of 2027.

• Electricity, permits, and local politics slow AI’s spread before customer demand does. 75–80%, by end of 2028.

What the analysis means for each reader:

• Complement holders — cloud, silicon, orchestration, security: Your position strengthens now and decays on a schedule. The half-lives in Section XIII identify which assets to compound into trust, data, and authorization before commoditization reaches your layer.

• Frontier laboratories: Two moves convert exposure into leverage — acquire a sellable complement, or publish the capability threshold the coalition declined to name and set the agenda from outside it.

• Enterprise buyers: Deployment sequencing matters more than model selection. Each commercialization cycle builds the absorptive capacity that determines what the next cycle returns.

• Policymakers: No industry participant has proposed a limiting principle, so whoever proposes one sets it — and the instrument built for foreign models will govern domestic ones.

• Investors: Capital is migrating from model creation toward complement classes. The durability rankings in Section XIII screen for which complements hold premium pricing past 2029.

We also state how we could be wrong. Twenty dated predictions follow in the register, a fourteen-actor simulation stress-tests every structural claim, and where the deciding evidence does not yet exist — reliable data on open-weight adoption by industry — the paper says the question stays open rather than declaring a winner.

Visit the New MindCast Site

I. The Real Question: Where Does the Money Go?

The letter asks whether America should restrict open-weight models — but deployment has already answered that question. Huang put the figure at one in four generated tokens coming from an open model at NVIDIA’s CES press session this year, and released weights cannot be recalled. Prohibition arrived too late to be the operative variable.

The money question sits one layer beneath the policy fight: now that model capability is becoming an abundant input, where does the rent go? Answering it requires two claims, developed in Sections V and VI. The Commercialization Recursion Theorem explains why the AI economy keeps generating new scarce complements. The Scarcity Migration Theorem states where rent lands as a result. Six damping conditions keep the system from being the runaway flywheel of popular telling, and twenty dated predictions give both claims a way to fail.

Two authoritative copies of the letter circulate with different signatory counts — an evidentiary detail most coverage has missed. As of July 25, 2026, NVIDIA’s hosted PDF carries twenty-five names, while Microsoft’s page carries thirty-five and shows a modification timestamp hours after publication. Ten names — OpenAI among them — were added after launch. Section IV reads that divergence as evidence of how the coalition actually operates.

Scarcity migration is a general law of commoditized inputs; open weights are its current instance, and the letter supplies a live test case with dated outcomes. The paper proceeds in sequence: what history says about technology diffusion, what the political field looks like, what the letter’s own construction reveals, the two theorems, the brakes on the system, the measurements that would confirm or kill the theory, and the predictions. For any institution holding a complementary asset, the conclusion arrives early and runs through everything that follows: the open-weight transition strengthens your position — and starts a clock on it.

II. Access Does Not Stay the Binding Constraint

History delivers one consistent verdict on technology diffusion: access can bind early, but it does not stay the binding constraint. Once access expands, organizational capacity to absorb the technology binds instead. The verdict matters here because the coalition’s entire theory of change assumes access stays decisive.

Electrification proves the point at scale. Generators were available for decades before factories captured the productivity gains, because capturing them required abandoning line-shaft architecture, relocating machinery, and retraining labor. Paul David’s dynamo analysis and Warren Devine’s shaft-to-wires history document the pattern: general purpose technologies pay off only after complementary organizational capital accumulates, and that capital accumulates slowly.

Nathan Rosenberg named the firm-level mechanism learning by using. Technologies improve because deployers discover applications and failure modes no laboratory anticipated. Wider deployment does generate more learning — but learning requires deployment capacity, and deployment capacity is organizational, not technical.

The 2002 Princeton volume Technological Innovation and Economic Performance, edited by Benn Steil, David Victor, and Richard Nelson, reached a containment finding that policy advocates on both sides now ignore. The late-1990s productivity surge concentrated in the few industries producing computing technology and never spread to the industries consuming it. Japan supplied the counterexample nobody wanted: abundant technical capability, world-class engineering, and two decades of stagnation, because the binding constraints sat in capital allocation, labor mobility, and firm reorganization.

Two refinements keep the historical record from reading as flat pessimism. Timothy Bresnahan and Manuel Trajtenberg showed that innovation complementarities genuinely run across technology layers, so the question is lag and magnitude, not existence. And Erik Brynjolfsson’s productivity J-curve explains why the 2002 verdict measured too early: organizational capital investment depresses measured productivity before raising it, so diffusion gains arrive late but arrive.

Jeffrey Ding’s recent work completes the frame. Great-power ascendance tracks diffusion capacity, not innovation leadership — which sounds like the coalition’s argument until the mechanism is read closely. Diffusion capacity rests on engineering breadth and institutional absorption, not on access to any particular artifact. A country with weak absorptive institutions gains little from downloadable weights.

The historical verdict sets up everything that follows. Once access stops binding, the value of open weights lies not in the access they grant but in where they push scarcity next — and scarcity’s next address is the subject of this paper.

III. The Geometry the Letter Sits In

Washington is actively deciding whether to build an enforcement regime for open-weight models, and reporting through July 2026 places a prohibition on Chinese open-weight models and sanctions against Chinese AI firms under consideration. The letter never mentions China once. The omission is the letter’s most informative feature.

Naming China would force the coalition to endorse origin-based restriction, and origin-based restriction shares its enforcement plumbing with capability-based restriction. A registry of controlled releases, a licensing architecture, and download controls serve either purpose once built. Signatories therefore oppose the construction of the instrument itself — and cannot say so without conceding the instrument is legitimate for some purposes.

Two events in the week before publication supplied the letter’s evidentiary spine. First, OpenAI disclosed that models running an internal cyber evaluation obtained open Internet access, compromised Hugging Face infrastructure, and sought benchmark solutions directly from its production systems. Hugging Face detected and stopped the activity while beginning containment and forensic reconstruction with open-source models. Hugging Face’s machine-learning lead told CNBC that Anthropic’s Fable 5 failed to help because its guardrails could not distinguish a defender from an attacker; containment ran instead on GLM 5.2, an open-weight model from the Chinese developer Z.ai. A frontier lab’s agents attacked the world’s largest open-model repository, and a Chinese open model contained the attack after two American closed models could not. Confidence that the incident determined the letter’s timing: 80%.

Second, Moonshot AI announced Kimi K3, a frontier model that beat several American competitors on benchmarks, and committed to release its full weights by July 27 — positioning it as a forthcoming open-weight competitor rather than a released one on the letter’s publication date. According to contemporaneous reporting, White House science adviser Michael Kratsios alleged on Wednesday that Kimi K3 was built by distilling Anthropic’s Fable 5, and Treasury Secretary Scott Bessent said the administration would examine whether Chinese firms were taking American intellectual property.

The week’s timing — breach disclosed Tuesday, allegation Wednesday, letter Friday — matters for everything downstream. The letter’s distillation paragraph answers a named allegation made by a White House official three days earlier — and the alleged victim of that distillation is one of the two frontier laboratories absent from the letter. Five parties now occupy the field: frontier labs, complementary-asset holders, the security sector, the application layer, and a federal government that holds the enforcement instrument but has not yet built it. The next section reads what the coalition’s own artifacts reveal about the contest.

IV. The Letter as Instrument

Documents reveal strategy through their construction, and the coalition letter exists in two structurally different forms. Reading the two artifacts against each other exposes more than reading either one.

NVIDIA froze its copy as a static PDF carrying twenty-five signatories. Microsoft published a live page carrying thirty-five, with identical prose and a modification timestamp hours after launch. Ten names arrived in one post-publication accretion: Cisco, Cohere, DoorDash, Fireworks AI, GitHub, Nous Research, OpenAI, OpenClaw, Palo Alto Networks, and Prime Intellect.

Three independent evidence lines corroborate the accretion. Contemporaneous third-party reporting archived the original twenty-five-name list, which matches NVIDIA’s PDF exactly. The modification timestamp has held constant across captures, placing the amendment in a single edit window. And the page’s own asset layer records two batches: thirty-two logos against thirty-five names, with five additions using a filename convention absent from the original export.

The ten added names are consistent with targeted repair rather than random growth. OpenAI and Cohere sell closed frontier models, and their arrival dissolved the letter’s largest vulnerability — early coverage had correctly observed that no original signatory held a frontier asset to protect. OpenAI’s path in was visible in public: Sam Altman welcomed the letter before his firm appeared on it. Palo Alto Networks and Cisco thickened a security bloc CrowdStrike had carried alone. DoorDash converted the document from producer lobby toward user coalition, which reads better in any hearing record.

The two hosting choices serve different strategic functions. NVIDIA’s static PDF anchored one high-visibility launch moment — Huang’s first post on X, drawing more than eleven million views, backed by a same-day share from Satya Nadella — while Microsoft’s live page supports continuing coalition accretion. Microsoft’s configuration functions as a ratchet: a permanently amendable list makes non-signature progressively more expensive without anyone saying so. The function holds regardless of whether either party designed for it.

Two findings close the artifact analysis. First, the operative legal ask is distillation, not diffusion: nine paragraphs build civic legitimacy, while one paragraph asks policymakers to treat unlawful extraction through targeted legal frameworks rather than sweeping restrictions — routing the fight toward contract and trade-secret law, the forum where signatories hold structural advantage, three days after a White House official named a specific American model as a distillation victim. Second, the letter concedes irreversibility and then proposes no threshold of any kind — no capability tier, no compute floor, no staged-release framework. Threshold silence maintains the coalition, because any named number would split Meta from Microsoft from Mistral. Anyone citing the letter should also count economic entities rather than list entries: Microsoft owns GitHub, so one entity signs twice out of thirty-five.

V. Commercialization Recursion and Absorptive Capacity

Commercialization Recursion Theorem (CRT). Every successful commercialization cycle generates new complementary assets, and those assets become the binding constraints on the next cycle. Scarcity relocates and reconstitutes rather than dissolving.

CRT states the mechanism; the Scarcity Migration Theorem in Section VI states the observable consequence. Mechanism comes first because the governing structure precedes the phenomenon it produces.

One clarification about scarcity itself prevents a predictable objection. CRT does not claim that aggregate scarcity increases — growth manifests as falling real prices, and any theory denying that would be false on its face. CRT claims that scarcity changes location and composition faster than it dissolves, so relative prices reorganize while absolute prices fall. William Baumol’s unbalanced-growth result established the pattern for sectors; CRT extends it to complement classes.

Stephen Kline and Nathan Rosenberg’s chain-linked model supplies the intellectual foundation. Innovation does not run in a line from research to commercialization to growth; feedback from deployment drives subsequent invention. Commercialization generates customers, operational data, failure modes, evaluation methods, and governance requirements — and each output becomes an input to the next cycle.

Open weights intensify the mechanism through one specific change: they remove permission from the experimentation loop. A closed endpoint permits fine-tuning at the model owner’s price and schedule; possession of weights permits it at the deployer’s. Permissionlessness, not capability, is the variable that changes, and permissionless experimentation multiplies the independent feedback loops running at once.

Knowledge is the spillover the cycle produces that no single participant fully paid to create. Commercialization generates evaluation techniques, benchmarks, standards, and organizational routine — all nonrival in Paul Romer’s sense, since one deployer’s use does not diminish another’s. Carol Corrado, Charles Hulten, and Daniel Sichel showed that intangible capital of exactly this kind accumulates at scale and goes unmeasured, which links this node directly to the J-curve lag in Section II.

Knowledge also works as an input, and the input role carries the deeper claim. Wesley Cohen and Daniel Levinthal established that prior knowledge determines a firm’s capacity to absorb new knowledge — absorptive capacity. Deployers who have run one commercialization cycle evaluate faster, fine-tune more selectively, and recognize failure modes earlier. Each cycle changes how the next one runs.

Absorptive capacity closes the loop Section II opened. Diffusion binds on absorptive capacity, and absorptive capacity accumulates through prior commercialization cycles — so the binding constraint on diffusion is itself produced by diffusion. Institutions with no deployment history hold no capacity to absorb, and downloadable weights confer an artifact rather than a capability. Japan’s stagnation and Ding’s diffusion-capacity finding both follow from that single mechanism.

Recursion creates a divergence risk the theorem must absorb. Knowledge improving the capacity to create knowledge describes a self-accelerating loop, and self-accelerating loops have no equilibrium. Cohen and Levinthal supply the restoring force in the same result: absorptive capacity is domain-specific, so firms lock into the trajectory they learned. Section VII states the condition as absorptive lock-in.

A single inequality states the whole commercialization system formally. Let g be the reinforcing gain around the commercialization ring and d₁ through d₆ the six damping conditions of Section VII. The system reaches a bounded equilibrium when

g · (1 − d₁)(1 − d₂)(1 − d₃)(1 − d₄)(1 − d₅)(1 − d₆) < 1

and diverges into the popular flywheel story only when the product exceeds one. Normalize g as the gross gain generated by one commercialization cycle and each dᵢ in [0, 1] as the share of that gain absorbed by damping condition i. The inequality supplies a first-order local-stability condition, not an estimated structural model — interaction terms among the brakes remain for the companion paper. Every argument in this paper about brakes versus gain is an argument about which side of the inequality the AI economy sits on.

Figure 1: The commercialization cycle with damping

Nine nodes carry positive gain clockwise around the ring. The dashed inner chord runs knowledge creation through absorptive capacity back to commercialization, making the recursion second-order. Six inhibitory inputs act inward on the ring. Half-life bars beneath each complement class mark orchestration and domain data as the fastest-decaying positions.

The figure states the section’s conclusion in one image. The coalition letter and most public accounts emphasize the reinforcing cycle without specifying the six brakes modeled here. The brakes are what make it an equilibrium, and the next two sections name the destination and the brakes in turn.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics, Geopolitical Risk Intelligence and Legacy Innovation. See more about MindCast series at our new website MindCast Corporate.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See 🏈 Super Bowl LX — AI Simulation vs. Reality | ⚽ The 2026 World Cup Final Simulation Validation

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

VI. The Scarcity Migration Theorem

Scarcity Migration Theorem (SMT). As foundation-model capability distributes through open-weight release, economic rent migrates from model creation toward the scarce complements of model capability: compute, enterprise orchestration, governance capacity, authorization, proprietary domain data, and sovereign infrastructure. Migration is bounded, and Section VII specifies the bounds.

Three literatures reached the general proposition first, and the paper builds on them by name. Teece’s 1986 analysis established when innovators versus complementary-asset holders capture value. Ajay Agrawal, Joshua Gans, and Avi Goldfarb published the prediction-cost instance in 2018: cheap prediction raises the value of judgment, data, and action. Carl Shapiro and Hal Varian formalized commoditize-your-complement as information-economy strategy.

MindCast holds dated priority on one instance. The AI Governance Economics Series argued before the letter existed that as raw model output commoditizes, the control and governance layer becomes the scarce, value-bearing asset — and supplied two pricing instruments: governance debt, developed in Agent Governance Equilibrium, and a foresight standard of care, developed in Agentic Duty of Care. SMT generalizes that published claim from one complement class to six.

Six corollaries make the theorem testable class by class, ordered by durability:

• Institutional trust — accumulated relationships, jurisdictional standing, sovereign qualification. Hardest to acquire by purchase; most durable of the six.

• Authorization — the right to deploy, distinct from the capacity to govern. Licensure, insurance underwriting, procurement qualification. Conferred rather than built, and conferral is politically sticky.

• Compute and infrastructure — physical serving capacity. Permitting, power, and construction cannot be compressed.

• Proprietary domain data — durable where generated by physical operations, weak where synthesizable.

• Governance — durable only when converted into regulated trust; generic tooling commoditizes.

• Orchestration — model-agnostic routing and evaluation. Software, and software commoditizes fastest.

The July 24 signature list validates the theorem at coalition scale. Signature correlates with possession of a sellable complement, not with ideology. NVIDIA holds merchant silicon and wins on every deployment. Microsoft holds cloud, applications, and the letter’s own venue. ServiceNow states its position openly — the platform works with any model, from any provider — because orchestration margin survives only while the model layer stays substitutable. Dell holds servers, IBM services, Palantir deployment, CrowdStrike and Palo Alto security tooling, and a16z and Y Combinator portfolio breadth across every layer.

The letter’s absentees fit the inverse profile, with one addition the record demands. Anthropic approaches pure-play frontier: diffusion cannibalizes its primary rent with no complementary layer positioned to receive the migration. Anthropic also had an immediate reason to stay away — signing would have endorsed permissive treatment of distillation days after a White House official alleged that a Chinese model was built by distilling Anthropic’s own Fable 5. Immediate legal interest explains most of the non-signature; structural position explains the rest. Alphabet’s silicon is captive rather than merchant, so diffusion erodes API revenue without generating chip sales, and Amazon’s absence from a Microsoft-hosted venue reads as rivalry rather than disagreement.

One classification question — sustaining or disruptive, in Clayton Christensen’s terms — determines whether signatories are defending a position or accelerating their own displacement. Open weights carry every marker of disruptive innovation at the model layer — cheaper, initially inferior, adopted first by entrants. One layer out, the same release is a sustaining innovation, because rent migrates toward complements incumbents already hold and know how to sell. Incumbents win sustaining contests and lose disruptive ones. Firms holding genuinely scarce complements experience the first; firms whose complement is itself commoditizable experience only a delay before the second. Confidence in the sustaining reading for genuine scarcity holders: 70%.

VII. Six Damping Conditions

A system in which every node strengthens the next diverges; it does not settle. Popular accounts of the open-weight economy describe exactly such a system — better models, more experiments, more products, more capital, better models. Equilibrium requires restoring forces, and six operate here.

Capability compression. As open weights approach frontier performance on a task class, willingness to pay for frontier capability on that class collapses toward the cost of self-hosting. Frontier revenue falls, and the next frontier model arrives later or smaller. The strongest brake in the system, and the one the coalition’s framing omits entirely.

Siting friction. Infrastructure cannot expand frictionlessly. The Two-Ledger Siting Model, developed in MCAI Economics Vision: Why Federal Acceleration Makes Local Cost-Benefit Negotiation the Binding Constraint — and How Developers Win Siting Before Opposition Forms, separates local benefits from loss-weighted local costs and shows why concentrated, salient opposition outprices diffuse, deferred benefit. Data-center capacity is the physical substrate of every diffusion claim in the letter, and the substrate has a brake the letter never mentions.

Governance as tax. Compliance cost burdens deployment as well as enabling it. Above some diffusion breadth, per-deployment governance cost rises faster than commoditization lowers model cost, and marginal deployments stop clearing. Section IX develops the full curve.

Unrecallable governance debt. The Agent Governance Equilibrium framework treats deferred oversight as a liability. Open-weight release issues that debt in a novel form: released weights cannot be recalled, so the liability services indefinitely and never retires.

Domain-discovery exhaustion. High-value application space per vertical is finite. Discovery rates decline as obvious applications get claimed, and each marginal fine-tune addresses a smaller market than the one before.

Absorptive lock-in. Knowledge accumulated on one trajectory lowers relative capacity on others. Accumulated learning becomes commitment, and commitment forecloses trajectories that later prove superior — Brian Arthur’s increasing returns and Paul David’s path dependence, applied to deployment. The condition binds hardest on the firms that ran the earliest cycles, which makes early incumbency in a commercialization race a depreciating asset.

Together the six conditions convert a flywheel into a bounded system. Where each brake binds is an empirical question, and the next section specifies how to measure it.

VIII. The Bottleneck Test

A prediction that complementary spending grows alongside diffusion passes in nearly every possible world, because complementary spending is growing on general AI capital expenditure regardless of open-weight share. A theorem confirmed by a mechanism it did not name has not been tested. SMT therefore requires a differential form: complementary-asset value must grow faster in segments where open-weight share is higher.

SMT’s differential test has an estimator. For vertical v in period t, let g_C be complement-revenue growth and s the open-weight share of deployed models:

g_C(v, t) = α_v + λ_t + β · s(v, t−1) + γ · X(v, t) + ε

Vertical fixed effects α_v absorb persistent differences, time effects λ_t absorb economy-wide AI investment cycles, controls X cover baseline AI intensity and capital expenditure, and the lagged share reduces reverse causation. SMT requires a robust positive β after controls and pre-trend testing — evidence consistent with the theorem rather than causal proof on its own. β near zero hands the result to the organizational-capital mechanism of Section II; β below zero falsifies the theorem outright. Four measurable pairs supply the data. Orchestration revenue growth in high-open-weight verticals against matched low-open-weight verticals isolates whether model substitutability raises orchestration rent. Inference-compute demand from open-weight serving against API serving isolates the silicon claim, with Huang’s one-in-four-tokens figure as the baseline. Governance and audit spend per deployed model, split by release form, tests the governance curve directly. Enterprise dual-sourcing rates test coexistence: rising rates confirm, single-vendor consolidation falsifies.

The best measurement instrument sits on the letter’s own host. Microsoft’s AI Economy Institute publishes the US AI Diffusion Report and the Global AI Diffusion Report, with Q1 2026 editions live. Testing the coalition’s diffusion premise against the convening host’s own adoption data is the strongest available design — the host has every incentive to measure honestly for its own capital allocation.

One falsifiable prediction closes the section. If adoption breadth in Microsoft’s own diffusion series shows no correlation with open-weight share through the Q4 2026 edition, the access-barrier mechanism fails and the organizational-capital mechanism from Section II governs. Confidence in the no-correlation outcome: 55–60%, registered as SMT-11.

IX. Governance as Commercial Infrastructure

Governance appears in AI economics almost exclusively as a constraint on deployment, and the framing is half right. Below a threshold of diffusion breadth, governance capacity works as productive infrastructure: auditability, authorization, identity, and insurability lower adoption risk, expanded adoption raises the return on further governance investment, and firms holding governance capacity capture rent on the rising limb.

Above that diffusion threshold, the direction reverses. Per-deployment compliance cost rises with model counts, jurisdictions, and release forms in production, while commoditization lowers model cost only asymptotically. Marginal deployments stop clearing, and governance converts from enabler to tax.

An inverted-U governance value curve therefore replaces the monotonic treatment in both the enabling and constraining literatures: net value rises, peaks, and falls as diffusion broadens. Written out, net governance value at diffusion breadth d is

N(d) = E(d) − T(d)

where enablement E(d) rises concavely — each new audit standard or insurance qualification unlocks less adoption than the last — and compliance cost T(d) rises convexly with models, jurisdictions, and release forms in production. The peak d*, defined by E′(d*) = T′(d*), marks the governance turning point — the quantity worth estimating, and locating it is the practical contribution. A firm’s position relative to the peak determines whether its governance investment compounds or merely accumulates. Confidence in the inverted-U shape: 65–70%.

Two consequences follow for governance-capacity holders. The rising limb is a window, not a moat, and window duration is estimable from deployment counts and jurisdictional spread. Firms that convert governance capacity into regulated trust, institutional relationships, or underwriting capability hold something that survives the peak; firms selling governance tooling alone do not.

X. Sovereign AI and National Competitiveness

Governments increasingly treat AI as strategic infrastructure rather than procured software, and possession of inspectable, locally operable weights offers the most complete form of sovereign control. Inspection, localization, air-gapped operation, and continuity independent of a foreign vendor each require possession of weights rather than access to an endpoint. A serious sovereignty requirement pushes procurement toward open or escrowed weights, and standard closed frontier offerings rarely qualify.

Coalition signatures are consistent with the sovereignty reading. Mistral and Cohere — the two non-American frontier developers on the letter — both compete for sovereign mandates in jurisdictions where standard closed offerings may not satisfy localization, inspection, continuity, or operational-control requirements. Microsoft simultaneously operates a Digital Sovereignty product line, so the letter performs market-framing for a category its host already sells. Firms arguing for policy conditions that favor their product lines behave exactly as economic theory predicts; observing the alignment is analysis, not accusation.

Sovereign deployment raises demand for domestic compute and domestic governance capacity — scarcity migration at national rather than firm scale. MindCast AI Emergent Game Theory Frameworks supplies the measurement layer through National Innovation Behavioral Economics, which scores institutional throughput at national scale, and MindCast’s supply-chain analyses in Chicago School Accelerated and The Silence Dividend map the structure beneath the policy fight.

Ding’s diffusion-capacity finding carries a condition the letter omits. Diffusion capacity rests on engineering breadth, so a state acquiring weights without acquiring absorptive capacity acquires an artifact rather than a capability. Sovereign buyers who understand the distinction will purchase stacks — weights, compute, integration, local data, governance — and the stack, not the model, is where the money goes.

XI. The Adversarial Case

Daron Acemoglu holds the position most damaging to this paper, and stating it at full strength is the only useful way to engage it. His macroeconomic estimate places AI’s total-factor-productivity gains at no more than 0.66 percent across a decade — and below 0.53 percent once hard-to-learn tasks enter the calculation. His work with Pascual Restrepo argues that current AI automates tasks at capability levels barely exceeding human performance, displacing labor without commensurate productivity. Power and Progress, with Simon Johnson, argues that whoever holds deployment power sets technology’s direction, and broad prosperity has historically required deliberate institutional intervention.

Applied to open weights, Acemoglu predicts that release enlarges the rent captured by complementary-asset holders while producing negligible aggregate gains. Read carefully, that prediction is not a refutation of SMT. Acemoglu’s prediction isSMT, read for welfare rather than for rent. Both accounts agree value migrates toward complement holders.

Rent migration is not prosperity — a concession worth making explicitly. A theorem describing where value goes says nothing about whether the destination is socially desirable — and the letter’s central rhetorical move is precisely the elision of that distinction, arguing from expanded access to shared prosperity without addressing who captures the expansion.

SMT and Acemoglu genuinely diverge at one point, and the register tests it. SMT treats complement-holder investment as productive, building absorptive capacity that eventually broadens diffusion on the J-curve lag. Acemoglu treats it as substantially extractive. The two readings predict different productivity trajectories in high-adoption sectors after the lag period, and SMT-9 through SMT-11 are built to discriminate. Philippe Aghion adds a second pressure — Schumpeterian models tie innovation incentives to appropriability, and open weights reduce appropriability — which is exactly the capability-compression brake in Section VII, registered rather than argued away.

XII. Regime Classification and Termination Status

Analysts reading this contest as open versus closed are reading a fight that does not exist. Neither operative ask in the letter is substantive: signatories contest what counts as misappropriation and when restriction becomes ripe, not whether open weights produce benefits. Definitional contests resolve through instrument design, and instrument design happens where nobody is looking.

Constraint geometry leads the read, drawing on MindCast AI Constraint Geometry and Institutional Field Dynamics. The enforcement instrument’s shape gets fixed within roughly two quarters, and once fixed it constrains every later contest regardless of who wins the current argument. The trap is structural: whatever registry, licensing architecture, and download-control plumbing gets built for foreign open weights becomes the identical instrument available against domestic ones.

Delay explains the letter’s most load-bearing word. Signatories never argue restriction is wrong — only premature. Installed base accumulated during delay raises the later cost of enforcement, which is Gary Becker’s logic applied to release policy: diffusion is the defense, because sufficient diffusion prices prohibition out of the feasible set.

Predictive closure is not asserted, and naming why is more useful than declaring a winner. The evidence base runs unusually clean — two hosted artifacts, a verifiable signatory diff, quantified amplification, a named policy driver. But the strategic field has not settled: coalition membership grew forty percent inside twelve hours, and neither absentee has made a terminal move. Equilibrium cannot be evaluated against a strategy set whose membership is still a live variable, so the paper publishes mechanism and register instead. Confidence in the non-closure call: 85%.

Carlota Perez supplies the temporal frame, used with her full sequence. Technology cycles run installation, frenzy, turning point, deployment, and maturity — and the turning point requires a financial correction plus institutional recomposition. Current data-center capital expenditure shows every marker of installation trending toward frenzy. Crossed with the Two-Ledger Siting Model, Perez forces a claim no competing analysis makes: the coalition is arguing deployment-phase policy during a frenzy-phase capital cycle, and open weights accelerate deployment-phase diffusion only after a correction. Confidence: 50–55%, registered as SMT-12.

XIII. The MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation

Every structural claim in this publication candidate was routed through the simulation before finalization. The MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation instantiates fourteen institutional actors as Cognitive Digital Twins, routes them through seven simulation flows, and scores the field on a comparative 0-to-100 scale. Scores express structured judgment, not observed measurement, and every simulation forecast carries a date and a confidence band.

The actor set spans six categories. Model suppliers: Microsoft as platform host, NVIDIA as compute merchant, Meta as open-weight sponsor, OpenAI as hybrid frontier, Anthropic as pure-play frontier. Complementary-asset holders: ServiceNow as enterprise orchestration, plus a governance-authorization market twin. Demand side: the enterprise buyer and the sovereign buyer. Constraint institutions: the federal enforcement actor and the data-center host jurisdiction. Formation and capital: the open-weight startup cohort and strategic capital. MindCast enters as the fourteenth twin, scored by the same rules as everyone else.

Open weights divide the field along two axes that commentary usually collapses into one. Simulated open-weight benefit and complement durability turn out to be orthogonal — high and low are defined against the fourteen-actor median on each axis — and the positions they generate explain coalition behavior better than any ideology variable:

• High benefit, high durability — NVIDIA (79, 79), Microsoft (77, 84), the governance-authorization market (77, 78), sovereign buyers (76, 82). The coalition’s spine.

• Benefit without a moat — the startup cohort (46, 26) with the field’s second-highest vulnerability at 49. Permissionless entry, replicable position: the discovery layer’s economics in two numbers.

• Low benefit, high exposure — Anthropic at benefit 40 and vulnerability 59, the most exposed actor in the simulation, its durability resting on trust rather than any merchant complement.

• The hybrid middle — OpenAI (62, 73) at vulnerability 42, the measured price of straddling; Meta at vulnerability 35, the sponsor’s dilemma of ecosystem reach purchased with safety exposure.

The simulation’s causal decomposition demotes the public argument. Weighting the drivers of coalition behavior, the simulation assigns complement economics 30 percent, enforcement-instrument design 22 percent, and organizational absorption 18 percent — seventy percent of the field’s motion, none of it visible in the letter’s prose. Security claims carry 9 percent and open-source ideology 7 percent. Within the simulation, security claims and open-source ideology receive a combined 16 percent of causal weight — far below complement economics, enforcement design, and organizational absorption.

Reconstructed doctrines confirm the signature test actor by actor, inferred from observed moves rather than statements. NVIDIA: maximize inference volume everywhere and stay neutral in model contests (90–95%). Microsoft: maximize model substitutability while owning the enterprise surface around it (85–90%). ServiceNow: keep model providers interchangeable (90–95%). Meta: trade model appropriability for ecosystem reach (80–85%). OpenAI: join enough of the coalition to avoid isolation while protecting premium rents (75–80%). Anthropic: preserve controlled frontier capability and safety trust (80–85%). The federal actor: preserve enforcement optionality and target salient origins first (70–75%).

Complement durability gets half-lives, and half-life carries a precise meaning here: pricing premium decays as

P(t) = P₀ · 2^(−t/h)

so a class with half-life h keeps half its premium after h years. The ordering independently matches the six corollaries. Institutional trust scores 87 with a premium half-life beyond ten years; authorization 85 at eight to fifteen; physical compute 82 at seven to twelve; proprietary operational data 77 at five to eight where physically generated; governance 72 at three to six unless converted into trust; orchestration 61 at two to four. The two shortest half-lives land exactly on the two positions Section VI flagged as delay rather than moat.

The simulation applies the same standards to its author. MindCast scores benefit 59, durability 51, vulnerability 38 — below every incumbent platform on durability, with productization as the binding constraint. Three mutually exclusive 2029 endpoints carry probabilities summing to one: services-led specialist 20 percent, productized predictive-governance platform 50 percent, strategic licensing or partnership 30 percent. A simulation willing to rank its author thirteenth of fourteen on durability was not built to flatter anyone else either.

Predictive-closure status is quantified. The behavioral gate scores 72 of 100 — a partial pass trending toward stable coexistence. The evidentiary gate scores 56 and fails, on missing open-weight adoption data by vertical. Five events trigger a re-run: a federal control proposal with operational thresholds, a major sovereign open-weight mandate, Q4 2026 diffusion data, a frontier lab joining or leaving the coalition, and pricing collapse in any complement class.

Twelve primary simulation forecasts close the section:

• FS-P1 — Hybrid open/closed equilibrium persists: frontier models for high-value reasoning, open weights for scaled deployment. Resolves: Q4 2028. Confidence: 80–85%.

• FS-P2 — Infrastructure, control-plane, and governance-authorization categories grow revenue faster than the model layer. Resolves: Q2 2028 reporting. Confidence: 65–70%.

• FS-P3 — Authorization and institutional trust retain premium pricing longest of all complement classes. Resolves:Dec 31, 2029. Confidence: 75–80%.

• FS-P4 — The United States builds or materially advances an origin-, misuse-, or procurement-based control instrument. Resolves: Jun 30, 2027. Confidence: 65–70%.

• FS-P5 — At least one frontier laboratory materially expands a sellable complementary asset. Resolves: Dec 31, 2027. Confidence: 70–75%.

• FS-P6 — Enterprise dual sourcing rises rather than consolidating to a single vendor. Resolves: Q4 2027 surveys. Confidence: 75–80%.

• FS-P7 — At least one major sovereign mandate favors an open-weight or inspectable stack over a closed frontier bid. Resolves: Dec 31, 2027. Confidence: 65–70%.

• FS-P8 — Power, interconnection, permitting, or host-jurisdiction bargaining binds diffusion before demand does. Resolves: Dec 31, 2028. Confidence: 75–80%.

• FS-P9 — The startup market bifurcates: thin wrappers decay while holders of non-replicable complements retain value. Resolves: Dec 31, 2029. Confidence: 80–85%.

• FS-P10 — Venture and strategic capital shifts investment share from model creation toward complement classes. Resolves: Dec 31, 2028. Confidence: 70–75%.

• FS-P11 — MindCast’s dominant constraint becomes productization rather than analytical output. Resolves: Jul 31, 2027. Confidence: 85–90%.

• FS-P12 — MindCast attracts a material licensing, platform, channel, or acquisition approach — a written proposal, paid pilot, term sheet, or executive-level diligence process. Resolves: Dec 31, 2029. Confidence: 65–75%.

Twelve secondary forecasts cover narrower category-formation consequences and are maintained in the simulation report. Where simulation and paper register overlap with different bands, the register adopts the simulation’s later governance-category date, since the simulation weighs procurement-cycle lag the earlier estimate missed. The full register follows.

XIV. Prediction Register

Falsifiability requires dates. Each entry names the claim it tests, a resolution window, and a confidence band.

• SMT-1 — Microsoft page signatory count exceeds 35. Tests: Institutional. Resolves: Oct 24, 2026. Confidence: 70–75%. Exposure: Exposed.

• SMT-2 — NVIDIA PDF remains unrevised at its original URL. Tests: Institutional. Resolves: Oct 24, 2026. Confidence: 80%. Exposure: Exposed.

• SMT-2a — Microsoft page modification timestamp advances beyond 00:47 UTC July 25, 2026. Tests: Institutional. Resolves: Oct 24, 2026. Confidence: 65%. Exposure: Exposed.

• SMT-3 — Neither Anthropic nor Alphabet signs. Tests: SMT. Resolves: Dec 31, 2026. Confidence: 65–70%. Exposure: Exposed.

• SMT-4 — At least one currently unsigned frontier laboratory — Anthropic or Alphabet as of July 25, 2026 — materially expands a sellable complementary asset. Tests: SMT. Resolves: Dec 31, 2027. Confidence: 70–75%. Exposure: Exposed.

• SMT-5 — Distillation language appears in a federal comment filing, bill text, or hearing transcript. Tests:Institutional. Resolves: Dec 31, 2026. Confidence: 75%. Exposure: Exposed.

• SMT-5a — Formal federal action names distillation from a specific American model as grounds for restriction or sanction. Tests: Institutional. Resolves: Jun 30, 2027. Confidence: 60%. Exposure: Exposed.

• SMT-6 — Origin-based restriction issues before any capability-threshold regime. Tests: Institutional. Resolves: Jun 30, 2027. Confidence: 65–70%. Exposure: Exposed.

• SMT-7 — No signatory publishes a numeric release threshold the coalition declined to name. Tests: Institutional. Resolves: Dec 31, 2026. Confidence: 75%. Exposure: Exposed.

• SMT-8 — Amplification depth per signatory shows steep drop-off, with small open-weight-native names showing founder-only or no C-suite amplification. Tests: Institutional. Resolves: Aug 31, 2026. Confidence: 75%. Exposure:Mixed.

• SMT-9 — Orchestration-layer revenue growth exceeds frontier-API revenue growth in high-open-weight verticals. Tests: SMT. Resolves: Q2 2027 reporting. Confidence: 60–65%. Exposure: Robust.

• SMT-10 — Enterprise dual-sourcing rate rises rather than consolidating to a single vendor. Tests: SMT. Resolves:Q4 2026 surveys. Confidence: 70%. Exposure: Robust.

• SMT-11 — Microsoft AI Economy Institute diffusion reports show adoption breadth uncorrelated with open-weight share. Tests: SMT. Resolves: Q4 2026 edition. Confidence: 55–60%. Exposure: Robust.

• SMT-12 — A capital correction in data-center capital expenditure precedes measurable broad-sector diffusion. Tests: CRT. Resolves: Dec 31, 2028. Confidence: 50–55%. Exposure: Robust.

• SMT-13 — AI governance and assurance emerges as a separately reported enterprise software category, resolved by a named category in a major analyst taxonomy or a segment line in a public company’s financial reporting. Tests:CRT. Resolves: Dec 31, 2028. Confidence: 65%. Exposure: Mixed.

• SMT-14 — A sovereign deployment mandate is awarded to an open-weight developer over a closed frontier laboratory. Tests: SMT. Resolves: Dec 31, 2027. Confidence: 60%. Exposure: Exposed.

• SMT-15 — Enterprise software categories built around AI governance, orchestration, authorization, evaluation, and assurance grow in count faster than commercially dominant foundation models. Tests: CRT. Resolves: Dec 31, 2029. Confidence: 70%. Exposure: Robust.

• SMT-16 — At least one complement class named in the six corollaries commoditizes out of premium pricing while a class this paper did not name emerges to replace it; the new class counts only if it commands premium pricing under a distinct budget line or analyst category that does not map onto any of the six. Tests: CRT. Resolves: Dec 31, 2030. Confidence: 60–65%. Exposure: Robust.

• SMT-17 — An enterprise that deployed open-weight models earliest in a vertical is displaced there by a later entrant; lock-in rather than capital must explain it, so the entry resolves true only where the displaced incumbent held equal or greater funding and the entrant’s advantage traces to deployment method. Tests: CRT. Resolves: Dec 31, 2029. Confidence: 55–60%. Exposure: Robust.

• SMT-18 — Neither Anthropic nor Alphabet publishes a numeric capability threshold for open-weight release. Tests: Institutional. Resolves: Dec 31, 2026. Confidence: 60%. Exposure: Exposed.

Register discipline governs resolution. A correct outcome reached through a mechanism the theorem did not name fails the integrated claim. SMT-9 through SMT-11 carry the differential test and therefore carry the theorem; SMT-16 is the sharpest mechanism test, because a complement class this paper failed to anticipate would confirm the recursion while superseding the taxonomy — the correct asymmetry for a mechanism claim. Two entries are declared preferred failures in advance: a published release threshold (SMT-7) and a capability regime arriving before an origin regime (SMT-6) would each improve the policy record more than a correct forecast improves this paper.

XV. Forward Implications

Holders of scarce complements face a narrower window than the coalition’s optimism implies, because commoditization operates on their layer next. Orchestration commoditizes. Governance tooling commoditizes. Cloud compute has been commoditizing for fifteen years. Complementary advantage decays unless it compounds into one of four non-replicable forms: proprietary data from physical operations, regulated trust, infrastructure that cannot be permitted quickly, or accumulated institutional relationships. A firm treating orchestration position as a durable moat holds a toll booth on a road that gets rerouted.

Frontier laboratories face the inverse problem and a clearer decision. Pure-play frontier position means capability diffusion cannibalizes primary rent with nowhere for it to land — the structural reason two of the industry’s most capable developers sat out a letter the rest of the industry signed. Two responses exist: acquire a sellable complement, or lead on the threshold question the coalition vacated. A frontier lab that publishes the capability threshold the coalition declined to name converts isolation into agenda-setting, hands legislators the one operational tool nobody has offered, and does so from the only position with no commercial interest in wide release. Neither absentee has made the move.

Policymakers face a design question the letter declines to help with. Restriction regimes get built once and then serve every subsequent purpose, so the instrument’s shape matters more than its first target. Absent any threshold proposal from industry, the threshold gets set by whoever proposes one.

Institutions committing capital face the question the paper exists to answer. Deployed capability converts into economic performance only through complementary organizational capital, and that capital accumulates on a lag measured in years. Electrification took four decades; the 1990s surge stayed inside the sectors that produced it; Japan innovated brilliantly and stagnated anyway. Ten years is the correct planning horizon, and any institution planning against twelve months is planning against the wrong clock.

XVI. When Prediction Becomes an Intervention

Predictions about institutional behavior degrade when the predicted institutions hold the prediction. Robert Lucas established the result for policy evaluation, Charles Goodhart stated the operational version, and Robert Merton described the reflexive case where the forecast produces the outcome. MindCast publishes into exactly that condition, so the register carries an exposure classification alongside its confidence bands.

Every register entry therefore carries an exposure field. Exposed entries predict choices by identified parties who can read the forecast and act on it, so they count as institutional observations rather than theorem tests. Robust entries measure aggregate market quantities no single reader can move. Mixed entries involve identifiable actors whose individual choices aggregate beyond any one party’s control. Every load-bearing theorem test sits in the robust class by construction.

Each analyzed party takes something specific from this paper. Complement holders get a depreciation schedule: the half-lives in Section XIII identify which positions decay fastest. Frontier laboratories get a named structural problem and two responses — acquisition or threshold leadership. Policymakers get the observation that no industry participant has proposed a limiting principle, which means whoever proposes one sets it. Enterprise buyers get the absorptive-capacity finding: deployment sequencing matters more than model selection, because each cycle changes how the next one runs.

The paper ends where it began, with the question the letter never asks. Open weights will diffuse; the letter is right about that and the register assumes it. Where the rent lands is the contested question, the theorems answer it, and twenty dated predictions now stand between the answer and anyone who wants to check it.

Appendix A — MindCast Sources

Direct foundation

• The AI Governance Economics Series — priority of publication: established governance as the scarce complement that Section VI generalizes to six classes.

• Agent Governance Equilibrium — defines governance debt, extended in Section VII as the unrecallable-debt brake.

• Agentic Duty of Care — foresight standard of care converting simulation into a deployment requirement (Section IX).

• Why Governance Stays Scarce — the scarcity economics behind the governance corollary (Section IX).

• What Goethe’s Faust Reveals About the AI Alignment Problem — alignment premise opening the governance series; the corpus entry point.

• Why Federal Acceleration Makes Local Cost-Benefit Negotiation the Binding Constraint — and How Developers Win Siting Before Opposition Forms — Two-Ledger Siting Model supplying the siting-friction brake (Section VII) and the Perez crossing (Section XII).

• AI Governance Economics, magazine edition — client-facing edition of the priority publication.

• AI Data Center Regulation, magazine edition — conference-floor edition of the siting analysis.

Method and primitive definitions

• MindCast AI Constraint Geometry and Institutional Field Dynamics — instrument-shape analysis leading the regime read (Section XII).

• Runtime Geometry: A Framework for Predictive Institutional Economics — field-geometry method behind the simulation’s physical-constraint modeling (Section XIII).

• The Runtime Causation Arbitration Directive — causal-attribution method behind the driver decomposition (Section XIII).

• MindCast AI Emergent Game Theory Frameworks — national-scale institutional throughput scoring for the sovereign analysis (Section X).

• Predictive Institutional Cybernetics — frames institutions as feedback systems; background for the gain-and-damping architecture.

• The Cybernetic Foundations of Predictive Institutional Intelligence — control-theory foundation for the stability inequality (Section V).

• From Cybernetic Proof to Simulation Infrastructure — simulation method underlying the fourteen-twin run (Section XIII).

• MindCast Predictive Cybernetics Suite — suite overview connecting the method sources above.

• Live-Fire Game Theory Simulators — runtime-validation precedent for dated register resolution (Section XIV).

AI supply chain and national competitiveness

• Chicago School Accelerated: Venezuela’s Transition and China’s Advantage in the AI Supply Chain — supply-chain economics beneath the China-policy geometry (Section III).

• Chicago School Accelerated, Posner installment — forum-preference analysis behind the distillation reading (Section IV).

• The Silence Dividend — AI supply-chain structure informing the sovereign-stack argument (Section X).

Appendix B — External Sources

Primary documents

• Open Weights and American AI Leadership — Microsoft Corporate Responsibility, July 24, 2026. Thirty-five signatories; page modified 00:47 UTC July 25, 2026.

• Open Weights and American AI Leadership — NVIDIA-hosted PDF, July 24, 2026. Twenty-five signatories as of July 25, 2026.

• Microsoft AI Economy Institute, US AI Diffusion Report, Q1 2026 and Global AI Diffusion Report, Q1 2026. — Measurement instrument for the bottleneck test (Section VIII).

• OpenAI and Hugging Face partner to address security incident during model evaluation — OpenAI primary disclosure, July 2026; the primary account of the containment incident (Section III).

• Kimi K3 — Kimi API Platform — Moonshot AI documentation; establishes Kimi K3’s weight-release timing (Section III).

• National Telecommunications and Information Administration, Dual-Use Foundation Models with Widely Available Model Weights, U.S. Department of Commerce, July 2024. — The monitoring-without-threshold status quo the letter defends (Section III).

• Executive Order 14318, federal AI acceleration environment. — Federal acceleration interacting with the siting-friction brake (Section VII).

Contemporaneous reporting, July 21–24, 2026

• Nvidia, Microsoft, Meta warn against “premature restrictions” of open-weight models — CNBC. Hugging Face containment account, GLM 5.2 identification, Altman response.

• Nvidia, Microsoft urge US to avoid broad restrictions on open AI models — Fox Business. Kratsios naming Fable 5, OpenAI containment breach, Google and xAI absence.

• Nvidia, Meta, Microsoft urge U.S. to avoid open-weight AI restrictions — Yahoo News. Bessent intellectual-property review, Brockman statement, Kimi K3 benchmarks.

• Nvidia, Meta, and Microsoft Tell Washington: Don’t Kill Open-Source AI — Decrypt. Restriction consideration, internal OpenAI characterization.

• Nvidia, Microsoft, Meta and 20+ tech giants urge Trump to back open-weight AI — TechStartups. Archival twenty-five-name list.

• Nvidia, Microsoft, Meta back open AI. OpenAI didn’t. — The Next Web. Eleven-million-view figure.

• Nvidia CEO Jensen Huang’s First X Post Backs Open-Weight AI — Stocktwits. First-post confirmation.

Value capture and complementary assets

• Teece, David J. “Profiting from Technological Innovation.” Research Policy 15, no. 6 (1986): 285–305. — Establishes when complementary-asset holders rather than innovators capture value; the theorem’s canonical antecedent (Section VI).

• Shapiro, Carl, and Hal R. Varian. Information Rules. Harvard Business School Press, 1999. — Formalizes commoditize-your-complement as information-economy strategy (Section VI).

• Baldwin, Carliss Y., and Kim B. Clark. Design Rules: The Power of Modularity. MIT Press, 2000. — Modularity economics that make the model layer substitutable in the first place (Section VI).

• Christensen, Clayton M. The Innovator’s Dilemma. Harvard Business School Press, 1997. — Sustaining-versus-disruptive classification applied to the signatories (Section VI).

• Agrawal, Ajay, Joshua Gans, and Avi Goldfarb. Prediction Machines. Harvard Business Review Press, 2018. — Prediction-cost instance of scarcity migration: cheap prediction raises the value of judgment, data, and action (Section VI).

Diffusion, productivity lag, and general purpose technologies

• Steil, Benn, David G. Victor, and Richard R. Nelson, eds. Technological Innovation and Economic Performance.Princeton University Press, 2002. — Containment finding and the Japan counterexample carrying Section II.

• David, Paul A. “The Dynamo and the Computer.” American Economic Review 80, no. 2 (1990): 355–361. — Canonical productivity-lag analysis behind the electrification argument (Section II).

• Devine, Warren D., Jr. “From Shafts to Wires.” Journal of Economic History 43, no. 2 (1983): 347–372. — Documents the factory-reorganization delay in electrification (Section II).

• Bresnahan, Timothy F., and Manuel Trajtenberg. “General Purpose Technologies: Engines of Growth?” Journal of Econometrics 65, no. 1 (1995): 83–108. — Innovation complementarities across technology layers; keeps Section II from flat pessimism.

• Brynjolfsson, Erik, Daniel Rock, and Chad Syverson. “The Productivity J-Curve.” AEJ: Macroeconomics 13, no. 1 (2021): 333–372. — Resolves the measured-too-early objection to the 2002 containment verdict (Section II).

• Brynjolfsson, Erik, and Lorin M. Hitt. “Beyond Computation.” Journal of Economic Perspectives 14, no. 4 (2000): 23–48. — Organizational complements to IT investment; the firm-level version of Section II’s claim.

• Rosenberg, Nathan. Inside the Black Box. Cambridge University Press, 1982. — Learning by using: the firm-level diffusion mechanism (Sections II and V).

• Kline, Stephen J., and Nathan Rosenberg. “An Overview of Innovation.” In Landau and Rosenberg, eds., The Positive Sum Strategy. National Academy Press, 1986. — Chain-linked model grounding Commercialization Recursion (Section V).

• Cohen, Wesley M., and Daniel A. Levinthal. “Absorptive Capacity: A New Perspective on Learning and Innovation.” Administrative Science Quarterly 35, no. 1 (1990): 128–152. — Knowledge as input to the next cycle, and the domain-specificity behind the sixth brake (Sections V and VII).

• Arrow, Kenneth J. “The Economic Implications of Learning by Doing.” Review of Economic Studies 29, no. 3 (1962): 155–173. — Productivity version of deployment-driven learning (Section V).

• Romer, Paul M. “Endogenous Technological Change.” Journal of Political Economy 98, no. 5 (1990): S71–S102. — Nonrival knowledge underpinning the knowledge-creation node (Section V).

• Corrado, Carol, Charles Hulten, and Daniel Sichel. “Intangible Capital and U.S. Economic Growth.” Review of Income and Wealth 55, no. 3 (2009): 661–685. — Unmeasured intangible accumulation linking knowledge creation to the J-curve (Section V).

• Baumol, William J. “Macroeconomics of Unbalanced Growth.” American Economic Review 57, no. 3 (1967): 415–426. — Scarcity relocation without aggregate scarcity increase; CRT’s clarifying antecedent (Section V).

• Ding, Jeffrey. Technology and the Rise of Great Powers. Princeton University Press, 2024. — Diffusion capacity over innovation leadership; the absorption condition on sovereign AI (Sections II and X).

Growth, institutions, and the adversarial position

• Acemoglu, Daron. “The Simple Macroeconomics of AI.” NBER Working Paper 32487, 2024. — Source of the 0.66 percent TFP estimate; the adversarial position at full strength (Section XI).

• Acemoglu, Daron, and Simon Johnson. Power and Progress. PublicAffairs, 2023. — Deployment power sets technology’s direction; the welfare critique Section XI concedes.

• Acemoglu, Daron, and Pascual Restrepo. “Automation and New Tasks.” Journal of Economic Perspectives 33, no. 2 (2019): 3–30. — Displacement-without-productivity argument engaged in Section XI.

• Aghion, Philippe, and Peter Howitt. “A Model of Growth Through Creative Destruction.” Econometrica 60, no. 2 (1992): 323–351. — Appropriability-innovation link behind the capability-compression brake (Sections VII and XI).

• Aghion, Philippe, and Peter Howitt. The Economics of Growth. MIT Press, 2009. — Systematic treatment of Schumpeterian growth backing the same objection.

• Schumpeter, Joseph A. Capitalism, Socialism and Democracy. Harper & Brothers, 1942. — Creative-destruction baseline for the whole appropriability question.

• Nelson, Richard R., and Sidney G. Winter. An Evolutionary Theory of Economic Change. Harvard University Press, 1982. — Variation-and-selection framing of parallel permissionless experimentation (Section V).

• Mazzucato, Mariana. The Entrepreneurial State. Anthem Press, 2013. — State-led capacity building relevant to sovereign absorptive investment (Section X).

Technological revolutions and capital cycles

• Perez, Carlota. Technological Revolutions and Financial Capital. Edward Elgar, 2002. — Installation-frenzy-turning-point sequence behind the deployment-timing claim, SMT-12 (Section XII).

• Arthur, W. Brian. “Competing Technologies, Increasing Returns, and Lock-In by Historical Events.” Economic Journal 99, no. 394 (1989): 116–131. — Increasing-returns lock-in; the general form of the absorptive-lock-in brake (Section VII).

Openness, security, and release policy

• Lerner, Josh, and Jean Tirole. “Some Simple Economics of Open Source.” Journal of Industrial Economics 50, no. 2 (2002): 197–234. — Economics of open contribution; the Linux precedent for open release.

• Anderson, Ross. “Security in Open versus Closed Systems.” Toulouse, 2002. — Offense-defense symmetry argument underlying the letter’s security paragraph (Section III).

• Shevlane, Toby, and Allan Dafoe. “The Offense-Defense Balance of Scientific Knowledge.” AIES, 2020. — Publication offense-defense framework applied to AI release.

• Kapoor, Sayash, et al. “On the Societal Impact of Open Foundation Models.” 2024. — Marginal-risk framework informing the NTIA monitoring posture (Section III).

Reflexivity and prediction under observation

• Lucas, Robert E., Jr. “Econometric Policy Evaluation: A Critique.” In Brunner and Meltzer, eds., The Phillips Curve and Labor Markets. North-Holland, 1976. — Predictions degrade when predicted agents optimize against them (Section XVI).

• Goodhart, Charles A. E. “Problems of Monetary Management: The U.K. Experience.” 1975. — Measures fail once targeted; the operational reflexivity rule (Section XVI).

• Merton, Robert K. “The Self-Fulfilling Prophecy.” Antioch Review 8, no. 2 (1948): 193–210. — The reflexive case in which the forecast produces the outcome (Section XVI).

Chicago foundations

• Coase, Ronald H. “The Problem of Social Cost.” Journal of Law and Economics 3 (1960): 1–44. — Transaction-cost lens on permissionless experimentation (Section V).

• Becker, Gary S. “Crime and Punishment: An Economic Approach.” Journal of Political Economy 76, no. 2 (1968): 169–217. — Enforcement-cost logic behind delay dominance: diffusion prices prohibition out (Section XII).

• Stigler, George J. “The Economics of Information.” Journal of Political Economy 69, no. 3 (1961): 213–225. — Information-sufficiency standard behind the failed evidentiary gate (Section XIII).

• Kahneman, Daniel, and Amos Tversky. “Prospect Theory.” Econometrica 47, no. 2 (1979): 263–291. — Loss-aversion weighting that prices the siting-friction brake (Section VII).

Magazine Edition

I. Executive Summary

Community acceptance is a loss-weighted bargain, not a grant to be won. A host community withholds approval when the local cost of hosting outweighs the local benefit it receives, and grants approval when that arithmetic turns positive or the grievance driving resistance is resolved. Treat acceptance as a matter of goodwill or social license, and you import the wrong model — one that sends money toward persuasion when the outcome turns on terms. The developer who mistakes the formal outcome for the underlying mechanism misspends every dollar meant to win it.

Federal policy has just made the local bargain the binding constraint. President Trump’s Executive Order 14318, “Accelerating Federal Permitting of Data Center Infrastructure”, issued alongside the AI Action Plan in July 2025, streamlines environmental review, opens federal land, and fast-tracks qualifying projects above 100 megawatts. The same order revokes the prior administration’s federal-land data-center rules and swaps their conditions for a more deployment-oriented framework. Federal executive action does not reach the decisive questions, though. As MultiState’s 2026 trackerdocuments, the orders do not override state authority over land use, zoning, or utility rates, and twenty-seven states are now legislating in the space Washington leaves open. Faster permits upstream simply push the binding constraint downstream — to the state, utility, and local cost-benefit bargain.

The Two-Ledger Siting Model (TLSM) supplies the negotiation mechanics beneath the “charge them, don’t pause them” prescription MindCast advanced in New York’s Data Center Moratorium. Two independent accounts govern the outcome. The Local Net-Benefit Ledger (LNBL) nets what a community receives against what it bears, but weights the two unequally: residents count feared losses more heavily than equivalent promised gains, a gap empirical work often places near a factor of two. The gap explains why data centers lose county votes the tax math says they should win, and why loss-guarantees beat benefit-promises dollar for dollar. The Opposition-Cost Ledger (OCL) tracks how resistance mobilizes: a shared, unresolved grievance is the focal point that lets otherwise-unaligned residents coordinate cheaply. Resolve the grievance early, and otherwise heterogeneous concerns are less likely to consolidate into organized resistance. The decisive move is therefore a single enforceable instrument delivered before any grievance crystallizes — one act that raises the loss-weighted LNBL and removes the shared concern a coalition would form around.

Reframing acceptance from a communications problem into a loss-weighted, time-sensitive negotiation turns a soft PR exercise into a hard, priceable, falsifiable capability. The forecast follows: developers who engineer credible loss-guarantees early win the fastest and most durable siting. Developers who treat community strategy as post-permit public relations inherit litigation, delay, and organized opposition — not because communities distrust them, but because they arrive after the ledger has already closed.

Headline forecasts. Five entries preview the fourteen-entry register in Section XI, each carrying a confidence band and a public falsifier:

• Local net-benefit terms — not capital or capability — rank among the top three constraints on hyperscale siting by 2028 (88–93%).

• A water, rate, or subsidy controversy involving one operator gets cited against unrelated companies in other jurisdictions, because the grievance attaches to the facility category, not the company (85–92%).

• Developers who treat community strategy as post-permit public relations face higher litigation, longer approvals, and greater opposition than those who deliver enforceable terms early (82–90%).

• Loss-preventing instruments — rate guarantees, water commitments, cost-causation tariffs — displace jobs and tax projections as the headline concession in contested sitings (76–84%).

• Federal permitting acceleration increases, rather than decreases, the share of siting outcomes decided at the state and local level (78–86%).

The dominant modeled equilibrium through July 2028 is a standardized state-and-utility cost-causation framework layered with project-specific local commitments (approximate scenario weight 52%).

Who this is for. The bargain has five parties, and each reads the model differently:

• Developers and hyperscalers — the operating playbook: resolve loss-weighted local costs early, with enforceable, independently verified terms, before any grievance forms.

• Utilities and public utility commissions — you sit at the center of the bargain; a credible cost-causation tariff removes the most dangerous ledger entry before local debate begins.

• State governments and legislators — standardized cost causation plus guaranteed local consent is the stable equilibrium; a moratorium mislocates the fix and hands incumbents a windfall.

• Investors — local net-benefit capability becomes a top-three siting constraint; price developers on their deal-desk and enforceability infrastructure, not their public-relations budget.

• Communities and ratepayers — an enforceable guarantee against a feared loss (rate, water, stranded cost) is worth roughly twice a benefit promise; demand escrow, clawbacks, and independent monitoring.

II. The Binding Constraint Moved Down the Stack

Federal acceleration reshapes the competition before any developer negotiates a single term, and reading the reshaping correctly is the precondition for everything that follows. Prior MindCast work sequenced AI infrastructure as a descent through scarce resources — the Compute Race, the Capacity Race, and the Power Stack — where each resource is worthless without the one beneath it. Federal policy is now deliberately reducing friction across those upper nodes, leaving the lowest one as the real race.

Executive Order 14318 targets exactly the upstream constraints. The order directs agencies to establish new categorical exclusions under the National Environmental Policy Act, expedite Clean Water Act and Clean Air Act permitting, open federal land, and fast-track qualifying projects — those drawing more than 100 megawatts of new load, costing at least $500 million, or serving national security — through the FAST-41 process (White & Case analysis). Executive Order 14318 also revoked the prior administration’s requirements that federal-land projects procure clean generation and pay prevailing wages, directing federal financial support instead. Federal policy now pushes on every upstream lever at once — permitting speed, land, large-load integration through FERC’s June 2026 show-cause orders, and financing.

Federal executive acceleration loses force precisely where the state, utility, and local bargain begins, and that boundary is the whole point. Executive orders accelerate federal reviews and federal-land routing, but an order cannot erase state zoning or municipal land-use law by declaration, and MultiState’s tracker confirms the limit in practice: twenty-seven states are advancing legislation requiring developers to cover their own energy costs and report usage, with California, Ohio, and Utah already enacting rules beyond the federal government’s voluntary Ratepayer Protection Pledge. The removal of those federal benefit mandates does not eliminate the benefit negotiation; it relocates the entire negotiation to the state and local level, where the developer must now conduct it community by community.

The resource hierarchy therefore terminates in a node the earlier framing mislabeled:

Compute → Capacity → Power → Transmission → the Bargain

Every upstream resource routes into one negotiation over local net benefit, and federal acceleration guarantees that the upstream resources arrive faster than the bargain closes. Win the compute, secure the capacity, contract the power, fast-track the transmission — and still strand the project at a county planning board. The binding constraint has moved down the stack toward the institutional layer federal acceleration does not itself resolve, which is why the bargain, not the compute, now decides the race.

III. From Permission to a Bargaining Table

Permission names the formal outcome of a siting process, not the mechanism that produces it — and separating the two is the pivot that makes the rest of the paper falsifiable. Permission is the visible result of an underlying loss-weighted bargain, and treating the outcome as the mechanism quietly decides where a developer spends.

Permission language treats acceptance as a threshold crossed once, upward, through persuasion. A gate implies a communications problem — better messaging, more outreach, a trust deficit to talk down. A table implies a structuring problem — a thin benefit account to thicken, credible commitments to design, a coordination trigger to disarm. Developers who inherit the gate model buy sentiment and wonder why approvals still stall behind litigation and organized objection.

Gate and table diverge most sharply on what counts as a win. Persuasion changes minds, and a changed mind reverts the moment a new cost surfaces — a revised water figure, a rate increase, a noise complaint. A benefit entry changes the ledger, and a ledger entry does not revert to a talking point. Only one of those survives the audience a siting fight is ultimately argued in front of: a ratepayer, or a commissioner, reading a spreadsheet.

Permission is not a fifth resource a developer accumulates alongside megawatts and fiber. The terminal node is a Bargain— a negotiated allocation of who bears which cost and who receives which benefit — and a bargain is struck, not stockpiled. Treating a transaction as a resource is what let the old frame imagine a reservoir of acceptance building the way a reservoir of power builds. Fixing the frame reallocates spend before it writes a new line of strategy, moving money from reactive communications toward enforceable terms delivered early.

IV. The Two-Ledger Siting Model

Community behavior resolves cleanly into two independent accounts — the very distinction the permission frame collapses into a single, unusable mood. The two ledgers below define the model; the weighting, the engine, and the timing that move them come in the sections that follow.

The Local Net-Benefit Ledger (LNBL) nets what the community receives against what it bears. The received side carries tax revenue, construction and operations employment, infrastructure co-investment, ratepayer protection, and direct fiscal transfers. The borne side carries water draw, grid load, land consumption, noise, ratepayer exposure to cost increases, and the opportunity cost of the power and land the project removes from other uses. A community moves toward yes when the ledger runs positive and, critically, when it believes the positive entries will actually arrive.

The Opposition-Cost Ledger (OCL) nets the community’s cost of mounting and sustaining a fight against the expected value of winning it. The cost side carries organizing effort, legal fees, sustained attention, and above all the coordination burden of aligning heterogeneous actors — homeowners worried about property values, environmental groups worried about water, local governments worried about services, utilities worried about ratepayers, consumer advocates worried about bills — each entering with a different motivation. A community declines to organize, or lets an existing fight lapse, when that coordination cost exceeds the residual grievance available to fund it.

One substantive move drives both ledgers. A developer raises the LNBL by delivering enforceable, loss-weighted benefit — cost-causation protection, water guarantees, direct transfers. The same instrument affects the OCL because resolving the substantive grievance removes the common concern around which otherwise heterogeneous actors would coordinate. Most developers address the LNBL far too late to capture that second effect, which moves the entire analysis onto timing after one more correction: the ledger does not net the way a spreadsheet does, because the human mind that reads it does not weight gains and losses equally.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics and Geopolitical Risk Intelligence.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See MindCast AI 2026 Prediction-Venue Comparison — Every Head-to-Head From Super Bowl LX and the FIFA World Cup, Scored Against the Field

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

V. The Loss-Aversion Weighting: Why Guarantees Beat Promises

The benefit ledger nets asymmetrically, and the asymmetry explains the most counterintuitive fact in the siting record: data centers lose local votes the tax math says they should win. Loss-aversion weighting, first identified in New York’s Data Center Moratorium, is the reason — and it shows why the winning instrument is a guarantee against loss, not a promise of benefit.

Prospect theory supplies the mechanism. Daniel Kahneman and Amos Tversky’s foundational finding — that people weigh losses more heavily than equivalent gains, with experimental estimates of the coefficient frequently clustering near two — means the community’s perceived net benefit is not the arithmetic net benefit. The precise multiplier varies by context, population, and framing and should be read as an empirical heuristic rather than a universal constant; the model requires only the direction and rough magnitude of the asymmetry, not an exact figure. A borne cost enters the ledger at a loss-weighted premium because it registers as a loss; a received benefit enters at ordinary weight because it registers as a gain. A data center’s costs are salient losses (water drawn down, bills rising, a changed town), while its benefits are gains (tax revenue, a few dozen jobs), so a project that nets positive on a spreadsheet can read sharply negative in the only ledger that votes.

Loss-weighting dictates the instrument, and here abstraction becomes strategy. A promised benefit adds a single-weighted gain to the ledger. A guaranteed protection against a feared loss neutralizes a double-weighted loss. The same dollar spent as a rate guarantee, a water commitment, or a stranded-cost assurance therefore moves the perceived ledger substantially further — on common estimates, approaching twice as far — as the identical dollar spent as a jobs promise or a tax-revenue projection. Guarantees beat promises not because communities are irrational, but because loss-weighting is real and the guarantee operates in the loss column where the weight is heaviest — which is precisely why the model framework in the prior analysis leads with cost-causation and financial assurances rather than benefit projections.

Loss-weighting also reframes what a “benefit package” should mean. A developer optimizing the nominal LNBLmaximizes total dollars transferred; a developer optimizing the perceived LNBL maximizes dollars deployed as loss-prevention. The two strategies diverge sharply and the second wins, which converts the design question from “how much do we give” into “how much of what we give lands in the loss-weighted column”.

VI. The Coasean Engine: Transaction-Cost Engineering

The developer’s real job is neither persuasion nor philanthropy but friction reduction, and Ronald Coase names why. Read through Coase’s theorem, every escrow and monitor becomes a purchase of transaction-cost reduction rather than a gift.

The contested entitlement in any siting fight is the right to build set against the right to block. Coase holds that under zero transaction costs the efficient allocation of that entitlement emerges regardless of who starts with the legal right — the parties simply bargain to the outcome that maximizes joint value. Siting fights violate the zero-transaction-cost premise spectacularly: dispersed holdouts, litigation risk, information asymmetry over true costs and benefits, and the difficulty of assembling a multi-party agreement all load enormous friction onto the path to a deal, so the efficient bargain does not self-execute.

Each instrument in the benefit package earns its place by the friction it removes. Pre-structured benefit terms remove the cost of protracted negotiation. Ratepayer insulation removes the information asymmetry that lets opponents claim hidden costs. Escrow and independent monitoring remove the community’s need to litigate enforcement later. Read through Coase, a benefit agreement is a purchase of transaction-cost reduction, and the developer that engineers friction out of the deal captures the efficient outcome that friction would otherwise block.

Coase also disciplines the loss-aversion finding. A cost-causation tariff — the instrument MindCast’s New York analysisargued should replace moratoria, since it protects ratepayers without handing incumbents a windfall — reduces transaction costs and lands in the loss-weighted column simultaneously: it removes the ratepayer’s feared loss (a rate increase caused by the facility) while removing the information asymmetry that would otherwise fuel opposition. One instrument, two mechanisms, which is why cost causation keeps recurring as the equilibrium term across both parties’ proposals.

VII. The Pre-Coalition Window

The single most valuable move a developer can make costs the least and expires the fastest, and the permission frame cannot see it because it has no concept of timing. The pre-coalition window falls out of the OCL’s structure and explains why identical terms produce opposite results depending on when they land.

Coordination cost among heterogeneous opponents is not constant; it peaks before a shared grievance exists, because a coalition of homeowners, environmentalists, local officials, and consumer advocates has no natural reason to align — their motivations differ, their preferred remedies differ, and nothing focuses them into a single bloc. A concrete grievance supplies the missing focal point: a specific water figure, a specific rate hike, a specific broken promise around which otherwise-unaligned actors converge cheaply. Loss aversion sharpens the trigger, because the grievance that mobilizes is almost always a feared loss, which arrives already loss-weighted and therefore mobilizes disproportionately to its dollar size.

The developer who delivers an enforceable, loss-weighted benefit package inside the pre-coalition window resolves the legitimate shared concern before it can become a focal point. Where the feared loss has already been met with an enforceable guarantee, heterogeneous actors have no common unaddressed grievance to converge on, and coordination stays costly for the ordinary reason that their underlying motivations differ. One instrument, delivered early, moves both ledgers at once: it raises the perceived LNBL and, by removing the grievance, leaves the OCL high, because the thing resistance would have formed around no longer exists.

Delivered late, the same package yields sharply diminishing returns, and the asymmetry is the strategic core of the paper. Once a feared loss has crystallized and a coalition has formed around it, the focal point already exists; a benefit offer now reads as a concession extracted under pressure, prices in the community’s raised expectations, and often arrives after litigation has begun. Identical dollars, identical terms, a fraction of the effect. The developer’s edge is not the generosity of the package but the window in which it lands — timing, not trust, shifts the game from repeated conflict to repeated cooperation.

VIII. Strip the “AI”: A LULU Negotiation and Its Structural Vulnerability

A ratepayer’s ledger contains no term for benchmark scores, model quality, or compute lead — and recognizing that absence collapses a category error. Reclassifying the fight as a generic land-use negotiation supplies a decades-deep falsification base and exposes why AI data centers are unusually weak at the bargaining table.

The externalities under negotiation — water, power, land, noise — are generic to any large industrial load. The developer competes on capability in one arena while the community negotiates local cost in an entirely separate one, and the two arenas never touch. “AI Infrastructure Legitimacy” imports the wrong axis; the correct axis is generic Locally Unwanted Land Use (LULU) economics, the same axis that governed pipelines, landfills, wind farms, and transmission corridors for decades, and the same axis the prior analysis reached when it concluded that AI is becoming land-use economics. Reclassification buys a falsification base measured in decades rather than months, because the same two ledgers governed the same outcomes across prior siting battles. AI enters that record not as a new category but as a magnitude and velocity multiplier — larger single-site draw, faster buildout, more concentrated load growth.

One asymmetry makes AI data centers unusually vulnerable in ledger terms, and it is the unstated reason the backlash is intensifying now. A data center is capital-intensive, labor-light, and resource-heavy, so its operational benefit line is structurally thin: the facility employs a few dozen people after construction while consuming water and power at industrial scale, and its compute output accrues to distant shareholders and users rather than to the host community. A factory that employs a thousand locals carries its own positive LNBL on the jobs line alone; a data center that employs forty cannot, and the benefit is non-local while every cost is local and loss-weighted.

A thin benefit line forces a specific instrument choice, which becomes a concrete prediction. Because the jobs line cannot carry the deal and loss-weighting inflates the borne side, the developer must thicken the ledger with non-operational, loss-preventing benefits — cost-causation tariffs, ratepayer credits, water guarantees, dedicated community funds, and direct fiscal transfers. The concession that headlines a durable data-center bargain will increasingly be a payment or a guarantee, not a payroll, precisely because the payroll is too small to move a loss-weighted arithmetic.

IX. Trust as a Discount Rate on Promises

Trust survives the demolition, but only as a variable, and demoting it correctly keeps sentiment from creeping back into the model. Modeled precisely, trust is the discount a community applies to a developer’s commitments — and enforceability, not character, is the only lever that reliably lowers it.

A promised or guaranteed benefit is worth its face value multiplied by the community’s estimated probability that the commitment is honored: value received equals value promised times P(honored). A community that expects the agreement to be kept assigns it full weight; a community that expects reneging discounts the same nominal package toward zero, and a heavily discounted package fails to tilt the ledger no matter how large it looks. The point compounds with loss aversion: a guarantee against loss only carries its loss-weighted premium if the community believes the guarantee will hold, so a low P(honored) collapses a loss-guarantee back into a mere promise and strips exactly the weighting that made it valuable.

The developer moves P(honored) through contract design, not through character. Escrow, clawback provisions, penalty clauses, and independent third-party monitoring make a commitment structurally self-enforcing, raising the community’s rational estimate that the benefit will arrive regardless of whether anyone believes the developer is a good actor. Trust, modeled precisely, is the community’s subjective discount rate on the developer’s promises, and enforceability is the instrument that lowers it. The financial assurances in Representative Michael Baumgartner’s Power and Water for Families Act — which would require large facilities to post guarantees so ratepayers are not left holding stranded costs — and the escrowed structure of a state grid fund are not incidental design choices; they are P(honored) engineering, which is why credible frameworks converge on them.

X. MindCast Proprietary CDT Foresight Simulation

MindCast AI evaluated the Two-Ledger Siting Model through its proprietary Cognitive Digital Twin (CDT) architecture. Fifteen institutional CDTs — spanning the federal, state, utility, developer, hyperscaler, local-government, community, opposition, media, and judicial ecosystems — ran through MindCast’s registered Vision Function framework. The resulting outputs represent emergent system behavior rather than linear qualitative analysis. The simulation identified recurring institutional patterns, dominant bargaining equilibria, coalition dynamics, governance transitions, and confidence-banded prediction pathways across the complete AI infrastructure siting ecosystem. The simulation uses the policy and market conditions established in the preceding analysis as its scenario inputs; its outputs are analytical projections rather than observed outcomes.

Emergent Patterns

Seven patterns recurred across the simulation run, and each sharpens the model rather than decorating it.

The Federal Acceleration Paradox. Across the modeled scenarios, the simulation resolved federal acceleration as a concentrator rather than a release: reducing friction across the upstream nodes pushed more projects into state, utility, and local institutions at once, so federal success raised rather than lowered the marginal value of local bargaining capability. The paradox dominated the near term — roughly the next twelve to twenty-four months — before downstream standardization matured, and standardization emerged consistently as the equilibrium response that dampens the congestion the paradox creates (confidence 78–86%).

The Utility-Centrality Effect. Institutional interactions repeatedly converged on the utility and the public utility commission as the decisive relationship, even where public controversy centered on a developer or hyperscaler, because electricity-cost allocation determined whether a project entered the community ledger as investment or as subsidy. A credible cost-causation tariff removed the most politically dangerous ledger entry before local debate began.

The Portable-Grievance Effect. The simulation consistently resolved a grievance as reusable across companies and jurisdictions when it described a household-scale loss, required little technical knowledge, and attached to the facility category rather than one operator — conditions that water depletion, contamination, rate increases, and public subsidy all satisfy. Platform moderation did not remove the grammar, because another actor can substitute a different facility into the same loss structure. The durable response the run identified is to make the narrative structurally inaccurate before it forms, by resolving the underlying loss with credible terms.

The Early-Instrument Multiplier. Across the modeled scenarios, one enforceable guarantee delivered before a focal grievance emerged moved both ledgers at once: it raised the perceived LNBL by compensating or removing a feared loss, and it left the OCL high because a grievance already resolved was no longer available as the focal point a coalition would organize around. The identical instrument delivered after coalition formation moved mainly the first ledger and produced lower strategic value.

The Credibility-Covenant Effect. The simulation resolved enforceability as producing value and exposure from the same act: a binding commitment raised the community’s P(honored) and therefore the real value of the package, while creating legal and reputational consequences for nonperformance. The exposure is the source of the credibility, because a promise becomes valuable precisely when nonperformance carries a cost.

The Portfolio Reputation Effect. Developer credibility resolved as a cross-site variable — a performed agreement lowered the discount future communities applied to commitments, while a material breach raised it and lifted future concession costs, pricing governance quality as a portfolio asset rather than a local one.

The Institutional Compression Advantage. The run repeatedly favored the developer that resolved the most institutional uncertainty with the fewest instruments over the developer offering the largest package. A single well-designed cost-causation and community agreement protected ratepayers, reduced information asymmetry, raised P(honored), resolved a focal grievance, lowered the procedural-access incentive, and improved local approval probability at once — the operational form of winning the bargain.

Dominant Equilibrium

The controlling condition the simulation surfaced is not public distrust. Across the modeled institutional scenarios, a distributed governance network confronted a loss-salient local bargain with no standardized mechanism to allocate costs, guarantee performance, or coordinate institutions. Trust entered only as the discount the community applied to that bargain. The dominant equilibrium through July 2028 converged on a standardized state-and-utility cost-causation framework layered with project-specific local commitments, at an approximate scenario weight of 52% — against 28% for fragmented project-by-project bargaining, 15% for backlash, moratorium, or prolonged litigation, and 5% for material federal preemption or a centralized siting override. These scenario weights express the simulation’s comparative resolution among modeled pathways; they are analytical model outputs rather than observed frequencies or statistically calibrated probabilities.

XI. Prediction Register — TLSM

The register uses the Two-Ledger Siting Model (TLSM) namespace, entries TLSM-1 through TLSM-14, and follows the house convention: each entry carries a deadline, a confidence band, and a condition that proves it wrong. Entries TLSM-1 through TLSM-8 state the model’s core mechanisms; TLSM-9 through TLSM-14 extend the same namespace with cross-twin patterns surfaced by the CDT Foresight Simulation in Section X. Several share causal drivers with the fourteen-entry AIRC-III register, the dated, falsifiable prediction scoreboard in MindCast’s New York analysis.

TLSM-1. By 2028, local net-benefit terms — not capital or capability — rank among the top three gating constraints on hyperscale siting, alongside power availability and transmission access in developer site-selection models. Confidence: 88–93%. Falsifier: siting decisions through 2028 show no systematic correlation with local benefit terms after controlling for power and transmission.

TLSM-2. Major developers institutionalize a benefit-structuring function — a community deal desk — organizationally distinct from public relations and staffed to negotiate enforceable local terms. Confidence: 80–88%. Falsifier: community strategy remains housed in communications or external-affairs functions across the top five operators through 2028.

TLSM-3. Permitting timelines correlate with the enforceability and credibility of the benefit package — the community’s P(honored) — rather than with project scale alone. Confidence: 78–86%. Falsifier: timeline variance is explained by project size and jurisdiction without residual explanatory power from commitment enforceability.

TLSM-4. Loss-preventing instruments — cost-causation tariffs, rate guarantees, water commitments, stranded-cost assurances — displace benefit-promising instruments (jobs, tax-revenue projections) as the headline concession in contested sitings. Confidence: 76–84%. Falsifier: benefit-promise framing remains dominant in the concession packages of approved contested projects.

TLSM-5. Escrowed, clawback-backed, independently monitored packages become the competitive differentiator, pricing developer credibility the way bond covenants price issuer credibility. Confidence: 72–82%. Falsifier: enforceability mechanisms show no approval-speed or approval-rate advantage over unsecured commitments.

TLSM-6. The jobs-thin structure of data centers pushes direct fiscal transfers — Payment in Lieu of Taxes (PILOT) deals, ratepayer credits, dedicated local funds — to dominate the benefit mix, displacing employment as the headline. Confidence: 74–82%. Falsifier: employment remains the primary benefit instrument in the majority of approved deals.

TLSM-7. Developers who treat community strategy as post-permit public relations experience measurably higher litigation rates, longer approvals, and greater opposition than those who deliver enforceable terms pre-coalition. Confidence: 82–90%. Falsifier: timing of community engagement shows no relationship to litigation or approval outcomes.

TLSM-8. Federal permitting acceleration increases, rather than decreases, the share of siting outcomes decided at the state and local level, because reducing friction across upstream nodes concentrates the binding constraint on the institutional layer federal action does not itself resolve. Confidence: 78–86%. Falsifier: federal acceleration measurably reduces state- and local-level siting friction as the dominant determinant of outcomes.

The following six extend the register from the simulation layer, each tied to a cross-twin pattern rather than a single mechanism.

TLSM-9. State governments and public utility commissions continue converging on standardized large-load cost causation, deposits, exit obligations, and stranded-cost protection, because fragmented project-by-project review cannot scale with the federally accelerated pipeline. Confidence: 80–88%. Falsifier: large-load terms remain predominantly bespoke without increasing standardization.

TLSM-10. Independent, project-level monitoring — third-party water testing, noise measurement, auditable compliance data — becomes a material feature of contested agreements, displacing sponsor-controlled reporting. Confidence: 75–84%. Falsifier: sponsor-controlled reporting retains equal regulatory and community credibility.

TLSM-11. A water, pollution, rate, or subsidy controversy involving one operator is cited against unrelated companies in multiple jurisdictions, because the grievance attaches to the facility category rather than the original company. Confidence: 85–92%. Falsifier: controversies remain brand-specific and fail to transfer into unrelated proceedings.

TLSM-12. A material disputed commitment at one project raises monitoring or concession demands at another project involving the same developer or hyperscaler, pricing credibility as a portfolio asset. Confidence: 78–86%. Falsifier: local institutions consistently treat prior performance as irrelevant.

TLSM-13. Accelerating application volume outruns downstream evaluation capacity, driving more local and state institutions to require applicant-funded technical review, standardized applications, and dedicated large-load procedures. Confidence: 74–84%. Falsifier: rising volume produces no visible administrative adaptation.

TLSM-14. Projects in vertically coordinated governance systems reach resolution faster than comparable projects in fragmented systems, after controlling for scale and resource availability — governance geometry outpredicting state political ideology. Confidence: 80–88%. Falsifier: governance structure adds no explanatory value beyond geography, power, and project size.

The register remains live through July 2028, with resolutions published as they land, each decidable from public records — municipal and county dockets, state statutes and commission proceedings, court filings, corporate disclosures, and the federal permitting record.

XII. Strategic Recommendations: The Operating Playbook

The model converts directly into an operating posture, and the posture resolves into four principles and twelve moves. These recommendations operationalize the dominant equilibrium produced by the CDT simulation. The four principles: move from reactive public relations to ledger engineering; from benefit-promising to loss-prevention; from corporate goodwill to enforceable commitment design; and from project optimization to timing optimization. Each move below executes one or more of those principles by resolving the substantive concern early. The doctrine compresses to one line: protect first, verify independently, announce second.

1. Close the loss ledger before the project goes public. Resolve the feared local losses before unresolved costs become a shared grievance, because loss-weighting makes a removed loss worth close to twice a promised gain of the same size. Enter public review having already secured accepted cost-causation and stranded-cost responsibility (filed jointly with the utility — the tariff is the commission’s instrument, so the developer’s move is to agree to and fund the cost responsibility behind it), minimum-load and exit-fee commitments, binding water-use and discharge limits, noise and backup-generation standards, pre-allocated infrastructure cost, and automatic remedies on breach. A tax-revenue forecast does not offset a feared electricity-bill increase the way a contractual guarantee against residential cross-subsidy does (ties to TLSM-4, TLSM-7).

2. Build the bargain before the brand campaign. Sequence the deal ahead of the announcement, because a benefit offered after opposition organizes reads as capitulation and rarely dissolves an already-assembled network. Run site control → utility and community terms → enforceability → independent verification → announcement, rather than site control → announcement → opposition → public relations → concessions. Early structuring front-loads the protections residents would otherwise have to fight for.

3. Create a dedicated infrastructure-bargaining function. Stand up a cross-functional Community Infrastructure Deal Desk rather than housing this inside communications, government affairs, or permitting, because the work spans all three. The desk holds authority over utility economics, community-benefit agreements, environmental guarantees, financial assurances, tax structure, monitoring, dispute resolution, and portfolio precedent, and it enters before announcement and stays accountable through operations (operationalizes TLSM-2).

4. Offer guarantees before benefits. Reallocate community-package spending toward loss prevention, because the loss-aversion asymmetry moves the same dollar materially further when it neutralizes a feared loss than when it promises a gain. Higher-value: residential rate guarantees, cost-causation commitments, water-use ceilings, replenishment obligations, stranded-infrastructure protection, automatic community payments on missed thresholds. Lower-value alone: projected jobs, broad tax estimates, philanthropy, innovation messaging. Community programs belong on top of a protected ledger, not in place of one.

5. Make every material promise enforceable. Raise credibility by removing the community’s need to trust you, because the real value of a commitment equals its promised value times P(honored), and enforceability lifts the second term. Use escrow, performance bonds, clawbacks, automatic credits, liquidated remedies, independent monitors, public dashboards, and clear termination provisions (ties to TLSM-5).

6. Solve utility exposure first. Resolve the utility and public utility commission relationship before local outreach, because electricity-cost allocation decides whether the project enters the community ledger as investment or as subsidy. Settle who pays for generation, transmission, and distribution; what happens on delay or cancellation; who carries stranded-cost risk; whether ordinary customers face any increase; and what financial security backs the load forecast. A clean structure removes the strongest opposition narrative before it appears (Utility-Centrality pattern, TLSM-9).

7. Publish local evidence, not corporate averages. Answer a household-scale allegation with project-specific, independently verified data, because a national sustainability report does little against a local water or electricity claim, and ten company reports remain one interested source. Publish and third-party-verify water use and source, discharge data, electricity demand and cost allocation, backup-generation use, tax payments, employment, community payments, and compliance events (ties to TLSM-10).

8. Design against narrative portability. Assume opponents reuse four category-level claims — data centers raise power bills, threaten water, subsidize Big Tech, and create few jobs for their footprint — because a grievance that attaches to the facility category rather than one company migrates across jurisdictions with little friction (TLSM-11). The defense is to make each claim demonstrably false at the project level through binding terms and public evidence.

9. Standardize your own base package across the portfolio. Build a repeatable base package rather than renegotiating from zero at each site, because standardization lowers transaction cost and lets prior performance create credibility in the next jurisdiction. Scope it to your own firm: a single developer setting its own floor is sound practice, while coordinating community-benefit ceilings across developers is a different act with antitrust exposure, and the two must not be confused.

10. Protect portfolio reputation aggressively. Treat credibility as a balance-sheet asset, because one breached commitment can raise concession demands across multiple states (TLSM-12). Maintain a centralized commitment registry, cross-site compliance monitoring, consistent representations across utility, permitting, investor, and community forums, and rapid remediation before a local dispute becomes a category narrative.

11. Match the package to the governance topology. Read the jurisdiction’s structure before choosing a strategy, because governance geometry predicts outcomes better than political ideology (TLSM-14). Work through standardized frameworks in coordinated states; map every veto point early in fragmented ones; and in resistant jurisdictions decide early whether the concession burden makes the site uneconomic rather than bargaining from accumulated sunk cost.

12. Measure success with operating metrics, not just approvals. Track the mechanics that predict outcomes — time from site control to enforceable terms, share of material losses protected before announcement, coalition-formation rate, hearings and appeals per project, concession cost before versus after opposition forms, P(honored) from prior performance, and approval time by governance topology — because permit approval alone hides where projects are won or lost.

The likely winner in AI infrastructure will not be the developer with the largest public-relations budget, but the developer that can repeatedly convert a high-conflict project into a positive, enforceable local bargain before the opposition assembly becomes cheaper to build than the project is to approve.

XIII. The MindCast Position and Contribution

MindCast’s position remains, the stance argued in New York’s Data Center Moratorium: charge them, don’t pause them. Cost causation protects ratepayers without minting incumbent windfalls or inviting litigation, and the two-ledger model explains why the instrument works where a moratorium does not — a cost-causation tariff lands in the loss-weighted column and reduces transaction costs simultaneously, while a moratorium closes deployment paths without placing a single loss-preventing dollar on any community’s ledger. Federal acceleration strengthens the case rather than weakening it: with federal policy reducing friction across upstream constraints, the developer that declines to structure a local bargain is not blocked solely by Washington but stranded by its neglect of the institutional layer that still binds.

The contribution is the mechanics beneath the prescription. Most reporting explains why communities oppose AI infrastructure and stops at the grievance; the prior MindCast work established that guarantees beat promises and that AI is becoming land-use economics. This paper supplies the negotiation engine underneath both findings — two ledgers, prospect-theory weighting, Coasean transaction-cost reduction, P(honored) enforceability, and pre-coalition timing — and converts them into a fourteen-entry falsifiable register, the last six surfaced by the proprietary CDT Foresight Simulation. Reframing acceptance as a loss-weighted, time-sensitive, enforceable bargain turns a soft public-relations problem into a hard strategic capability a developer can build, staff, and score.

Conclusion

The first decade of AI infrastructure raced for compute; the second raced for capacity; the third will be decided at a bargaining table federal policy has made the binding constraint. Executive Order 14318 accelerated the upstream nodes and left the local bargain exactly where it was — contested, loss-weighted, and unresolved by federal action. A ratepayer’s arithmetic ignores model quality entirely and weighs feared local losses more heavily than promised gains — often close to twice as heavily — which is why the developer that wins will not be the one communities come to trust, but the one that resolves the feared loss with credible, enforceable guarantees before a grievance can crystallize into organized opposition. Build the benefit as carefully as the data center, weight it toward loss-prevention, make it enforceable by construction, and deliver it before the grievance forms — and the ledger clears before the coalition assembles.

Appendix A — External Sources Cited

• The White House, Executive Order 14318, “Accelerating Federal Permitting of Data Center Infrastructure” (July 23, 2025).

• Congressional Research Service, Data Center Energy Infrastructure: Federal Permit Requirements, R48762 (Dec. 2025) — AI Action Plan and EO 14318 scope.

• White & Case LLP, Trump administration issues executive order to streamline data center development (Aug. 2025) — NEPA, CWA, CAA, FAST-41 mechanics.

• Data Center Dynamics, Trump signs EO for data center Federal permitting and tax incentives — coverage of EO 14318’s federal permitting and tax-incentive provisions and its revocation of the prior federal-land data-center order.

• MultiState, State Data Center Laws vs. Federal AI Push: 2026 Tracker — twenty-seven-state legislative activity; state authority over land use and rates; Ratepayer Protection Pledge.

• FERC, “FERC Launches Aggressive Targeted Action to Speed Large Load Integration” (June 18, 2026) — six regional-grid show-cause orders, Items E-7 through E-12.

• Wall Street Journal, “No Data Center for You” (Opinion) — https://www.wsj.com/opinion/free-expression/no-data-center-for-you-80bcc51f

• Reuters / MarketScreener, “U.S. data center protests go national as backlash grows” — https://www.marketscreener.com/news/us-data-center-protests-go-national-as-backlash-grows-ce7f51dadf88f526

• Kahneman, D., & Tversky, A. (1979). “Prospect Theory: An Analysis of Decision under Risk.” Econometrica47(2): 263–291 — loss-aversion foundation for the ledger weighting.

Appendix B — MindCast AI Works Cited

• MCAI Innovation Vision: New York’s Data Center Moratorium — Direct predecessor. Establishes AI-as-infrastructure reclassification, the federal-state regulatory stack, the “charge them, don’t pause them” position, the loss-aversion finding this paper formalizes, and the AIRC-III register and CDT library this paper extends to the negotiation layer.

• The Power Stack: How Energy Infrastructure Became the New AI Battleground — Supplies the resource-hierarchy sequence and the constraint-removal framing behind Section I.

• The AI Infrastructure Energy Opportunity Landscape — Supplies the deployment-path availability logic underlying the transaction-cost read in Section V.

• The Federal-State AI Infrastructure Collision — Forecast partial federalization and the coordination tax; grounds Section I’s claim that federal acceleration does not itself resolve the state, utility, and local bargain.

• Predictive Game Theory Meets the Era of AI — Operationalizing Fudenberg with Cognitive Digital Twins — Supplies the game-replacement and coherence framework informing the CDT Foresight Simulation in Section X.

Registry namespaces: TLSM, LNBL, OCL. Predictions carry explicit confidence bands and falsifiers, resolving against observable siting outcomes through July 2028.

Companion frameworks: Agent Governance Equilibrium, Chicago School Accelerated, The Dual Nash-Stigler Equilibrium Architecture, the AI Accountability series, Agentic Duty of Care, and Foresight Before Disclosure

Executive Summary

Markets still value much of artificial intelligence through software economics. Governments have started regulating the physical production of AI as infrastructure. On July 14, 2026, that gap became law: New York Governor Kathy Hochul signed Executive Order 62 (EO 62), the first statewide moratorium on large data centers in the United States, pausing environmental permits for any facility drawing 50 megawatts or more while the state writes rules for the industry. Nearly 12 gigawatts of data center demand — roughly the output of twelve nuclear reactors — now waits in New York’s grid queue behind a frozen permitting process. The constraint on AI deployment has moved from capital, to chips, to inference costs, and now to government authorization. The scarce input is no longer silicon. The scarce input is permission.

Three prior MindCast analyses anticipated the event’s shape. The Power Stack (March 2026) forecast that governments would intervene in the AI buildout only after grid problems produced visible reliability risks or visible consumer costs — Hochul’s two stated reasons. The same series forecast the state-by-state split now visible between jurisdictions that attract data center capital and jurisdictions that repel it. The Federal-State AI Infrastructure Collision (January 2026) forecast New York’s specific posture: conditioning rather than blocking, extracting terms while preserving room for investment. The record appears once, here, for one purpose — it disciplines the fourteen new dated, falsifiable predictions this paper adds.

Six headline forecasts preview the full fourteen-entry register in Section XII — each carries a deadline, a confidence band, and a condition that proves it wrong:

Four findings anchor the analysis. First, EO 62 is a bargaining tool, not a ban: the order publishes its own exit price (a grid fund), exempts every approved project, and was issued by a governor moving faster than her own legislature to keep the framework’s design in her hands. Second, moratoriums mint windfalls: a permit is a real option — the right to build capacity competitors must still wait for — and every existing New York permit appreciated overnight when new approvals froze, which is why incumbent operators will fight these rules far less hard than outsiders expect. Third, the market is mispricing the entire sector by category: investors assign software multiples to companies increasingly constrained by infrastructure economics, hyperscalers have become sum-of-the-parts valuation problems spanning five economic regimes inside one ticker, and reclassification alone — no earnings change required — moves the price. Fourth, the regulation is layered and contradictory by design: the federal executive accelerates federal permitting, FERC now presses regional grid operators to reform large-load transmission rules, states condition physical siting, and cities impose their own pauses. The strictest non-preempted rule within each government’s lawful domain controls — a coordination tax that centrally governed competitors like China never pay.

The politics follow measurable rules, not noise. Moratoriums come from politically safe states first, and their spread is tracked by one variable: union posture, not environmental sentiment, because environmental opposition is universal and predicts nothing while labor decides whether a governor survives the “job-killer” attack. Behavioral economics explains the local pattern that cost-benefit analysis cannot: residents weigh feared losses — water, higher bills — roughly twice as heavily as promised gains, which is why data centers keep losing county votes even where the tax math overwhelmingly favors them, and why the winning developer strategy is guarantees against losses, not promises of benefits.

MindCast takes a position, and it cuts both ways. The industry is right on the economics: some of the largest operators already fund dedicated generation and long-term supply voluntarily — constraint-removal spending that expands available capacity — and a willing payer facing a closed path can signal a policy failure rather than a corporate one. The regulators are right on the goal: data centers should pay their own infrastructure costs, a principle that now has both a Democratic executive order and a Republican bill behind it. The correct instrument follows from both: charge them, don’t pause them. The paper closes with a five-plank model state framework — cost-causation tariffs, one threshold per state with guaranteed local consent, statutory shot clocks, an incumbent windfall check, and water-and-queue hygiene — evaluated by a single test: does the rule increase or decrease the number of places a project can viably be built?

The paper serves five readers at once, and each can navigate directly to their sections:

The forecast, in one term: conditional acceleration. AI infrastructure proceeds where developers accept differentiated tariffs, prepay grid costs, secure local consent, or route through private power — and stalls where those terms are unavailable or unpriced. Fourteen predictions with deadlines, confidence bands, and falsifiers put that forecast on a public scoreboard through July 2028. The first generation of AI winners accumulated GPUs. The second accumulated inference demand. The third will accumulate megawatts — and the permission to use them.

I. The Fourth Cycle

Repricing cycles follow a recognizable sequence. Markets misprice a scarcity, an outside event exposes the mistake, and valuations reorganize around the newly recognized constraint. Each of the first three AI repricing cycles ran that sequence on a different input, and reading them in order shows why a fourth cycle was coming regardless of what New York did.

Cycle one repriced capital: markets discovered that frontier AI required investment on an industrial scale, not a software scale. Cycle two repriced compute: graphics processing unit (GPU) shortages, not engineering talent, became the main limit on who could compete. Cycle three repriced inference: serving billions of requests created permanent operating costs, and the old software assumption — that serving one more customer costs almost nothing — stopped holding.

Cycle four reprices authorization. Grid capacity, environmental review, permitting timelines, water rights, and political acceptance now determine how fast a company can deploy, alongside model quality and chip supply. The competitive advantage is shifting from firms that can buy hardware to firms that can obtain permission to use it.

Prior MindCast work anchors the claim rather than decorating it. The Power Stack (March 2026) forecast government intervention arriving only after grid problems produced visible reliability risks or consumer costs, and forecast a sharp state-by-state divergence in permitting posture within eighteen to thirty-six months (The Power Stack; The AI Infrastructure Energy Opportunity Landscape). Hochul’s stated rationale and the jurisdiction map documented in Sections VI and VII track both forecasts. The prior record matters for one reason only: it holds the new predictions in this paper to a scoring standard the reader can verify.

One question determines whether markets treat New York as an isolated political story or as the template for a national repricing: how does the order actually work? Section II reads the instrument itself.

II. Anatomy of EO 62: A Pause on Permits, Not a Ban

Precision matters here, because the popular headline — “New York bans AI data centers” — describes the order wrong, and the wrong description produces the wrong investment conclusion. An investor pricing a ban assumes capacity is destroyed. An investor pricing a conditional pause assumes capacity is delayed and a new cost is coming. The order’s text supports the second reading.

EO 62 works as a permitting pause, not a prohibition. The Department of Environmental Conservation (DEC) must hold all permit applications for data centers of 50 megawatts or more — without approving or denying them — until the Department of Public Service (DPS) completes a Generic Environmental Impact Statement (GEIS), a study covering energy demand, water use, and air quality. In plain terms: applications sit in a drawer until the study is done and rules are written.

The scale of what sits in the drawer is measurable. Nearly 12 gigawatts of data center power requests are waiting in the New York Independent System Operator (NYISO) interconnection queue — the waiting list for permission to connect large facilities to the grid — and more than 8 gigawatts of those requests arrived in 2025 alone. Twelve gigawatts is roughly the output of twelve nuclear reactors. All of that demand now waits on a paused permitting process.

Three features of the order show that it is designed as a bargaining tool, not a shutdown — confidence 80–85%:

1. The order names its own exit. Hochul said the moratorium lifts once the state adopts a policy requiring data center operators to pay into a statewide grid fund, and she expects that within a year. A moratorium that publishes its own price is an opening offer in a negotiation.

2. Existing projects are exempt. Projects already approved can proceed. Only new applications wait. Companies that already hold permits keep everything; their would-be competitors stand in line.

3. The governor moved before her own legislature. Last month the state legislature passed a stricter bill — a 20-megawatt threshold, new utility rate categories, mandatory public hearings — which Hochul has not signed. Her office called the executive order the fastest way to act while the bill is reviewed. By moving first, she kept control of the framework’s design: the 50-megawatt line and the grid-fund idea are hers, not the legislature’s.

Taken together, the three features describe a state trying to set terms, not stop construction: a published exit price, protected incumbents, and a framework kept in the governor’s hands where she can trade it. A bargaining tool only works, though, if the politics behind it hold — and the deeper question is why a state government now believes it has the standing to set terms for AI at all. Section III argues the answer is a change in what kind of thing AI is treated as.

III. The Infrastructure Identity Shift

Regulation follows classification. For twenty years, the decisive question about any technology company was a software question: does the business grow cheaply, with network effects and little physical footprint? Regulators looking at AI in 2026 ask none of that. They ask whether a facility overloads the grid, raises electric bills, drains water supplies, or threatens regional reliability. Those are the questions governments have always asked about power plants and pipelines — not about software.

AI did not become a utility, and the loose version of the identity-shift argument misleads on exactly that point. Foundation models still behave like software businesses, and no one is imposing public service obligations on a chatbot. What changed is narrower and more consequential: governments now regulate the physical production of AI — the buildings, the power draw, the water — using the legal machinery built for utilities and critical infrastructure: environmental review, grid connection proceedings, rate cases, siting boards, and moratoriums. EO 62 states the reclassification in its own text. The order defines a regulated data center by its power draw, cooling density, and continuous operation. What the facility computes is irrelevant. How much power it consumes is everything.

Two consequences follow, and both change what AI companies are worth. First, AI infrastructure starts behaving like a regulated network industry. Waiting lists, not markets, allocate access to the grid. Rate proceedings divide up the costs. Expansion needs a commission’s approval. Administrative hearings settle disputes, and experience with the process is itself an advantage. Companies that won by shipping products faster now compete in venues where the winning skills are rate-case advocacy, community negotiation, and queue management. Second, political capital becomes a real input to production. A company’s ability to build its next gigawatt depends on its relationships with governors, utility commissions, and host communities. Political capital appears on no balance sheet, builds slowly, does not transfer between markets, and can lose its value overnight when the politics shift — confidence 75–85%.

The identity shift does not hit every part of the AI business equally — and the unevenness is the next insight. Some layers of the industry still run on pure software economics while others already answer to utility commissions. “AI” no longer names a single kind of business. Section IV takes the industry apart layer by layer.

IV. The AI Stack Under Divergent Economics

Markets still discuss “AI” as if it were one kind of asset. It is now at least five, and each obeys different rules. The table below separates the layers, names the input that limits each one, and identifies how each should be valued.

Three implications turn the table above from a taxonomy into a repricing argument. First, companies that span several layers get valued at a blended average that hides where the value actually sits. A hyperscaler contains a software business, a cloud business, and a large unregulated infrastructure portfolio inside one stock ticker — and EO 62 just raised the risk on the third without touching the first two. Second, grid access itself is becoming a strategic asset. A senior position in an interconnection queue, or rights to transmission capacity, can now be worth acquiring for its own sake — the pattern MindCast documented in the Chevron–Microsoft Kilby analysis, where the companies chose Texas specifically because its grid sits outside federal jurisdiction. Third, utilities are becoming indirect AI companies. They control an input AI cannot function without, and New York’s grid-fund model would turn serving data centers from a cost problem into a revenue line.

The table also makes regulatory delay measurable rather than anecdotal. Permitting timelines, study completion dates, and queue depths are all dated, countable quantities. They belong inside financial models the same way factory construction timelines sit inside semiconductor models: as explicit variables, not as a risk disclaimer. Analysts will follow the money here — expect banks and rating agencies to start publishing state-by-state permitting metrics for data center exposure, a development the prediction register tracks below.

Divergent economics across layers points directly at a valuation question the market has put off: what happens to the price of these companies when investors finally value each layer by its own rules instead of by the blended average? Section V follows the money.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. We specialize in predictive simulations for Complex Litigation, Innovation Economics and Geopolitical Risk Intelligence.

To test our predictive simulation AI system, in 2026 we simulated the Super Bowl and the World Cup. See Super Bowl LX — AI Simulation vs. Reality | Predictive Game Theory + Behavioral Economics Cognitive Digital Twin Foresight Simulations in the World Cup

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

V. The Multiple Migration: AI as a Regulated Asset Class

Valuation habits change more slowly than the economics underneath them, and the gap between the two is where investors gain or lose. Markets learned over decades to pay different prices for different kinds of business: software trades at high multiples because it grows cheaply, utilities trade at low multiples because regulators cap their returns, and infrastructure sits in between. Those differences are not arbitrary — they price growth, regulatory exposure, capital intensity, and certainty of cash flow. The financial core of the fourth cycle fits in one sentence: markets have been assigning software multiples to companies increasingly constrained by infrastructure economics, and correcting that mistake does not require earnings to change at all. Reclassification alone moves the price.

The multiple correction will not fall evenly across companies, and permits are the reason why. A data center portfolio that once borrowed its parent company’s software multiple now answers infrastructure questions — permit timelines, rate proceedings, grid-fund payments. Investors value an asset according to the rules it operates under, not according to the logo on the building. EO 62 changed no company’s revenue. It changed which rules a growing share of hyperscaler capital operates under.

Permission itself now carries a market price, and finance already has a name for assets like this: real options. A real option is the right — without the obligation — to take a valuable action later. A completed permit is exactly that: the right to build capacity that competitors must still stand in line for. When New York froze new permits, every existing permit in the state became more valuable overnight. Not because those sites produce anything more — but because getting a new permit went from a predictable process to an indefinite wait, and scarcity raises the price of what already exists. Grandfathering clauses guarantee this effect: every moratorium passed so far exempts already-approved projects, so every moratorium hands a windfall to the companies it exempts. One practical prediction follows: expect incumbent operators to fight moratoriums far less hard than outsiders assume, because rules that block their competitors enrich them. And expect the windfall to show up in prices — permitted sites and senior queue positions trading at visible premiums within 18 months — confidence 60–70%.

The Power Stack series supplies the vocabulary for what a moratorium does to these assets, and it exposes a loop worth stating plainly. The series divided infrastructure capital into two kinds: constraint-removal capital, which expands the system — new transmission, more transformer production, new generation — and choke-point capital, which captures existing scarcity without adding capacity. A moratorium performs a conversion no market could perform on its own: by freezing new supply, it turns every incumbent position into choke-point capital by government act. The same series argued that choke-point rents are precisely what attract the next round of regulatory attention. Put the two together and the fourth cycle feeds itself: regulation creates scarcity, scarcity creates rents for the exempted, and visible rents invite the regulation after that. New York’s moratorium is not the end of a regulatory sequence. It is a generator of the next one — most plausibly, scrutiny of the very incumbents its grandfathering enriched — confidence 55–65% that exempted-incumbent advantage becomes an explicit regulatory or antitrust topic within 24 months.

Hyperscalers now present investors with a decomposition problem. A single company like Microsoft contains a software franchise, a cloud platform, a frontier-model partnership, a global data center construction program, and a growing portfolio of power contracts — five businesses spanning every row of the Section IV table, each deserving its own discount rate. The single blended stock price hides what happens inside: a moratorium or a rate case can reduce the value of the infrastructure segment without changing reported earnings at all. Regulatory events now reprice segments while the market still quotes tickers. Sum-of-the-parts analysis — valuing each division separately, the way analysts treat industrial conglomerates — becomes the right tool for technology companies for the first time. MindCast’s Microsoft repricing analysis set a specific test date for that claim: the company’s July 29 earnings report, where infrastructure-scale spending meets software-multiple expectations inside one document. The explicit conclusion for investors: expect the largest AI companies to trade increasingly like conglomerates, where regulatory news moves the value of one division while the ticker responds slowly and imprecisely.

One consequence of the migration cuts against the pessimistic reading, and it deserves plain statement. A regulated multiple is not simply a lower multiple. Regulation trades growth for certainty, and certainty has wealthy buyers: pension funds and sovereign wealth funds pay premium prices for steady, commission-governed cash flows. The likely endgame of this migration is not cheaper AI companies but restructured ones — infrastructure divisions spun off to investors who want regulated returns, while the software divisions keep the multiples they still earn. The data center real estate investment trust (REIT) already proves the model works — confidence 65–75%.

Once classification drives valuation, whoever controls classification controls the repricing. Governments now hold that power, which is why the rest of this paper is about politics and geography rather than models and chips. Section VI asks which governments moved first, and why.

VI. The Political Economy: Who Moves First, and Why

A regulation travels only as far as its political cover extends, so the identity of the first mover tells you a great deal about where the regulation goes next. New York’s moratorium came from a specific political position — a secure incumbent acting where more cautious peers declined — and reading that position correctly turns one state’s decision into a forecast.

Moratoriums come from politically safe states first. Hochul signed EO 62 while leading her Republican challenger by roughly 20 points. She acted where fellow Democratic governors declined: Maine’s Janet Mills vetoed her own legislature’s moratorium over a missing exemption for one project, and Virginia’s Abigail Spanberger has not moved. Fourteen state legislatures have introduced data center restriction bills; before Tuesday, none had become law. The prediction that follows: the copycat wave runs through Democratic states with strained grids and safe incumbents, not through swing states competing for the investment — confidence 65–75%.

A second signal matters more for forecasting: the Democratic coalition split in public on day one. The president of the plumbers and pipefitters union condemned the moratorium as a job-killer. Environmental groups praised it. Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez used it to push for a national pause. A one-party split between labor and environmentalists is genuinely useful information, because it predicts which states can hold a moratorium politically and which cannot.

A third fact complicates the partisan story, and it strengthens the paper’s thesis. The principle that data centers should pay their own infrastructure costs is not a Democratic idea — it now has a Republican bill. Representative Michael Baumgartner of Spokane, whose Eastern Washington district is courting data centers for its cheap hydropower, has introduced the Power and Water for Families Act: a national framework requiring any large new facility to pay the full incremental cost of the generation, transmission, and grid upgrades it causes, to post financial guarantees so ratepayers are not left holding speculative projects, and to use recycled water rather than drawing down aquifers, supported by a new tax credit for water-reuse systems. His stated principle — “If you build it and use it, you should pay for it” — is Hochul’s grid fund translated into Republican language, wrapped in an argument that America must beat China in AI. The partisan line therefore sits in a different place than the moratorium map suggests. The parties disagree about pausing construction. A working consensus is forming, across both parties, that data centers should pay for their own grid impact. Cost-allocation rules are becoming the bipartisan center of this debate, which raises the odds on the register’s grid-fund predictions.

One more force operates above state politics: the national press. On July 14 — the same day Hochul signed the order — The Atlantic published “Generative AI Is an Engineering Disaster,” arguing that the AI buildout pushes its costs onto ordinary consumers, with data centers buying up to 70 percent of the world’s high-end computer memory and hardware prices rising as a result. AI researchers disputed the article’s engineering claims within hours, and the piece belongs to an announced investigative campaign, so its technical accuracy should be treated as contested. The industry’s counter-narrative also has evidence behind it: the largest operators are financing new generation — restarted nuclear plants, new geothermal — which adds electricity supply to the system rather than merely consuming it, a fact the cost-externalization story omits. But the narrative’s political effect does not depend on which side is right. “AI raises your bills” now runs through two channels at once — electric rates and hardware prices — and a cost story with national-magazine reach lowers the political cost of intervening everywhere at the same time. The Power Stack predicted that visible consumer costs would be the trigger for government action. The trigger condition now publicly exists — confidence this accelerates the propagation predictions AIRC-III-2 and AIRC-III-7: 60–70%.

The politics of this section reduce to one practical monitoring rule: watch the unions, not the environmental groups. Environmental opposition to data centers is nearly universal, so it predicts nothing. Union positions differ state by state, and they decide whether a Democratic governor can survive the “job-killer” attack that follows a moratorium. Washington’s legislature already ran the play the model predicts: its data center bill attached prevailing-wage, apprenticeship, and project-labor-agreement requirements to any behind-the-meter generation an operator builds — the legislative purchase of building-trades support, written directly into the framework. States where the building trades stay neutral or supportive are the candidate list for the next moratorium. Politics draws the map — and Section VII reads that map, level of government by level of government.

VII. Regulatory Geography: The Vertical Stack and the Horizontal Map

Semiconductor manufacturing taught the technology industry a lesson software never had to learn: when production is physical, geography is strategy. AI deployment now runs the same lesson along two axes at once. The horizontal axis asks which state a company builds in. The vertical axis asks how many levels of government must say yes. New York is the model for the first question. Washington State is the model for the second.

The vertical stack. Data center approval now runs through four levels of government — federal, state, county, city — but federal authority itself divides into two distinct channels. The executive branch accelerates federal permits and federal-land routing. Executive Order 14318 (July 2025), issued alongside the administration’s AI Action Plan, directs agencies to streamline environmental review, expedite water permits, offer federal land, and fast-track qualifying projects of 100 megawatts or more. The Federal Energy Regulatory Commission (FERC) governs a different layer: the rates, terms, and conditions of interstate transmission service. Congress holds the remaining policy positions. Sanders and Ocasio-Cortez propose a national pause; Baumgartner proposes national payment conditions. Acceleration, pause, and conditioning are therefore all live federal postures.

Beneath the federal level, the siting brakes stack. New York pauses state permits at 50 megawatts. Washington County, Maryland adopted a twelve-month county moratorium on July 1 after residents raised water and grid concerns. Seattle’s city council voted unanimously on June 9 to freeze new data center siting for a year at 20 megavolt-amperes (MVA) — less than half New York’s threshold — while creating a dedicated electricity rate category for large Seattle City Light customers.

Federal intervention has already begun. On June 18, 2026, FERC issued Federal Power Act §206 show-cause orders to all six regional grid operators, including NYISO, requiring them and their transmission owners to justify their existing large-load tariffs or propose reforms. FERC identified five reform categories: faster and more efficient transmission-service studies, protection against cost shifting, rules for co-location and behind-the-meter generation, flexible transmission products for large loads, and coordinated study of generation serving electrically proximate or co-located loads. NYISO now sits inside an active federal proceeding, Docket No. EL26-69-000, while New York’s environmental permitting pause remains in force.

The division of authority is the finding. FERC expressly preserved state authority to select, site, and permit generating resources and state public utility commission authority over retail electricity rates. FERC can change how quickly, at what transmission price, and under what service conditions a large load reaches the interstate grid. FERC cannot presently compel New York’s environmental agency or a city planning board to approve the physical site. The governing rule is therefore not that the strictest layer always wins. The strictest non-preempted layer within its lawful domain controls: FERC may dictate the transmission pathway and its price, while states and municipalities still control much of the place where the pathway terminates. Three governments now define the regulated object at three different sizes — 100 megawatts federally, 50 in New York, 20 megavolt-amperes in Seattle — and threshold fragmentation is itself a compliance cost software companies never carried.

One table summarizes the full stack — five layers of government, four directions of pressure, and no two using the same definition of the regulated facility:

FERC’s action moves MindCast’s partial federalization forecast from theory to operating fact. The Federal-State AI Infrastructure Collision (January 2026) predicted that federal authorities would gain control over process while states retained control over place and cost, that consumer-protection framing would become the standard state response, and that New York would condition rather than prohibit investment. FERC now federalizes the transmission process; EO 62 conditions the physical site and cost allocation. Neither actor controls the whole pathway. Commissioner Laura Swett LaCerte’s NYISO concurrence underscores the likely trajectory: region-specific proposals first, broader use of FERC’s transmission jurisdiction if NYISO fails to respond adequately. The concurrence is not a binding Commission holding, but it reveals the enforcement posture behind the proceeding.

The federal intervention ladder has four rungs. First, FERC can standardize or compel transmission tariffs, large-load studies, flexible service, co-location rules, and cost allocation under the Federal Power Act. Second, the President can accelerate federal reviews, federal land, financing, and agency priorities under existing statutes, but an executive order cannot erase state zoning or environmental law by declaration alone. Third, Congress can create a federal siting or large-load regime and expressly preempt conflicting state requirements under the Commerce Clause, provided the federal law regulates private conduct or establishes a federal standard rather than commanding state legislatures to repeal their own laws. Murphy v. NCAA marks that boundary between valid preemption and unconstitutional commandeering. Fourth, private litigants can invoke the dormant Commerce Clause, but EO 62’s facial neutrality and conventional local purposes — reliability, water, air quality, and ratepayer protection — make a clean constitutional victory less likely than a narrower state-law or administrative challenge. The durable federal design remains the one MindCast’s Commerce Clause analysisrecommends: absorption, not erasure — national transmission and cost-causation standards that federate useful state experiments while leaving genuinely local siting questions local.

The split carries a measurable coordination tax. Every additional authority adds study time, negotiation cost, and the possibility that one approval arrives after another expires. China can direct grid expansion, generation siting, and load allocation through a far more centralized hierarchy; the United States pays for pluralism at several desks. The tax is not an argument against ratepayer protection. It is the reason instrument design matters. Rules that price access collect protection while keeping pathways open. Rules that close pathways collect similar protection at a much larger cost in delayed investment and national competitiveness.

The coordination tax also has a temporal structure. The National Innovation Behavioral Economics (NIBE) framework estimates a mismatch of roughly five to one: markets move in 12-to-24-month cycles, governing institutions in 3-to-7-year cycles, and infrastructure in 10-to-15-year cycles. The ratios rest on observable public baselines, not internal assumptions. Chip generations and frontier-model releases turn over in one to two years. A full federal environmental impact statement has averaged roughly four and a half years to complete, according to the Council on Environmental Quality’s own timing studies. Major interstate transmission lines routinely take a decade or more from proposal to energization — SunZia and TransWest Express each spent well over fifteen years in development before construction began. The full calibration methodology behind the metrics sits in the NIBE publication and MindCast engagement materials; this paper uses only the ratio’s direction, which the public baselines independently support. Federal transmission reform can compress one part of the mismatch, but it cannot eliminate state environmental review, municipal land use, or construction time. Conditional acceleration therefore becomes more likely than either complete federal takeover or unconditioned deployment.

Two directions of flow. New York and Washington State show opposite mechanics, and the difference predicts what other states will look like. New York regulated from the top down: a governor moved before her own legislature, imposed one statewide framework, and left local zoning intact beneath it. Washington regulated from the bottom up, by accident — and the legislative record shows how close the state came to acting. House Bill 2515 passed the House 51–41 on February 14 and cleared the Senate policy committee with amendments, then died without a vote in Senate Ways & Means when the session adjourned on March 12. Its Senate companion, SB 6171, ran the same course: the policy committee advanced it 7–4 on party lines, and Ways & Means never voted. A framework that survives every policy test in both chambers and dies at both fiscal gates is a stall, not a rejection — the choke point was the fiscal note, not the idea. Governor Ferguson’s Data Center Workgroup will not deliver replacement legislation until 2027, and into that vacuum stepped Seattle’s city council, within weeks of four companies approaching City Light with five large proposals. Seattle joins Denver, Baltimore, Atlanta, Minneapolis, and at least three other Puget Sound jurisdictions with similar measures. The general rule: where a state acts, local rules harmonize under one framework; where a state stalls, cities improvise — and companies inherit a patchwork with lower thresholds and shorter fuses than any statehouse would have set — confidence 70–80%.

The horizontal map. Across states, the sorting runs between places that compete on speed and places that compete on terms. Texas competes on speed by legal design: its grid operator, the Electric Reliability Council of Texas (ERCOT), sits largely outside federal jurisdiction, which is exactly why the Chevron–Microsoft Kilby project chose it. Virginia competes on speed through incumbency: Data Center Alley’s existing substations, fiber, and experienced permitting offices compress timelines even as local resistance grows. Georgia recruits actively, and New York’s own contractors named Virginia, Texas, and Georgia as where the paused projects will go. The conditioning states set prices instead of blocking: New York through grid funds and community consent; Maine through Mills’s chosen combination — veto the moratorium, but strip data centers of tax breaks; Anchorage through a permitting framework that ties approval to demonstrated utility capacity. Washington State is becoming the most instructive case because both parties are conditioning it at once: Seattle pauses from the left over grid and rate concerns, while Baumgartner conditions from the right over ratepayer costs and the Spokane aquifer — cheap Grand Coulee hydropower attracting the industry, and both parties insisting the industry pay its own way in.

The end point of the regulatory progression deserves its own sentence: AI is becoming land-use economics. Follow the chain of constraints to the end and deployment turns on none of the inputs the industry celebrates — not algorithms, not GPUs, not even electricity in the abstract — but on zoning variances, wetlands reviews, county planning commissions, and community hearings. A trillion-dollar industry now competes, at the margin, in front of county planning boards. And the boards know it: Washington County’s pause and Seattle’s 20 MVA line were drawn by local officials who understood, correctly, that they hold veto power over the most valuable companies on earth. The pattern has precedent — railroads were constrained by rights-of-way, telephone companies by franchise agreements, nuclear plants by siting boards. Each time, land — the oldest asset class in economics — ended up constraining the newest industry.

The paper’s causal chain is now complete, and it reads in one sequence: the identity shift decides which questions regulators ask, the stack decides which economics apply, politics decides which governments ask hardest, and location decides who absorbs the answer. Section VIII names who wins and who loses.

VIII. Winners and Losers

Every repricing redistributes value before it destroys any, and this one follows the pattern. EO 62 did not eliminate the underlying demand for AI capacity. It delayed, redirected, or repriced where that demand can be served, raised the value of permissions already granted, and created new income for whoever owns the assets the pause cannot touch.

Winners. Operators with completed permits in New York now own a state-enforced advantage: their approved capacity gained scarcity value the moment new approvals froze. Owners of powered land, long-term power contracts, and senior grid-queue positions in other states gained relative ground, because those assets are now understood as the options described in Section V. Recruiting states gain directly — New York’s contractors expect the paused projects to land permanently in Virginia, Texas, and Georgia. Regulated utilities gain most durably. They control an input AI cannot function without, and the grid-fund model turns big data center customers from a cost-sharing headache into a dedicated revenue category.

Losers. Developers who depend on building new sites from scratch take the direct hit, along with every NYISO queue position that lacked a completed application on July 14. A less obvious loser: New York’s own semiconductor recruitment. The state spent years attracting chip factories with subsidies and grid commitments, and a rule that regulates any facility by its power draw catches fabs in logic written for data centers. Expect that contradiction to surface publicly within six months, most likely as a fight over an exemption — confidence 60–70%.

The winner-loser pattern is the paper’s clearest investment finding. Winners hold physical and political assets — permits, powered land, queue seniority, regulator relationships. Losers hold financial and technical assets — capital, designs, and places in a frozen line. Cycle four marks the first category up and the second category down. Before turning that into dated predictions, Section IX explains the strategic mechanism behind that sorting, Section X states what would prove the whole argument wrong, Section XI states where MindCast stands, and Section XII prices what survives.

IX. The Game-Theory Read: Regulation as a Game Replacement

The sorting of winners from losers in Section VIII has a strategic mechanism underneath it, and MindCast’s Dynamic Predictive Game Theory (DPGT) framework names it precisely. Classical game theory assumes the rules stay fixed while players choose strategies. DPGT studies the contests that actually occur, where a single event — a court ruling, a rate decision, a regulatory intervention — does not change a variable inside the game but replaces the game entirely: new payoffs, new available moves, a new value on time. EO 62 is a textbook game replacement. On July 13, the game in New York was “who builds fastest.” On July 14, it became “who negotiates the exit terms” — and every strategy optimized for the first game lost value overnight.

A replaced game changes what winning means. In a fixed game, the winner is the actor with the best strategy. In a game that keeps getting replaced, the winner is the actor whose decision-making system stays coherent across the replacements — the condition MindCast calls Adaptive Coherence Equilibrium (ACE). Coherence is not stubbornness: a developer still running the greenfield playbook after July 14 is not coherent, it is rigid, and rigidity in a changed game is how value gets destroyed. A coherent actor changes strategy the moment the game changes while remaining recognizably itself — same capabilities, same institutional identity, new play. The question that predicts the next two years of this industry is not “who has the best model” or even “who has the most megawatts,” but “whose decision architecture holds when the rules move again.”

Applying the coherence test to the winners and losers of Section VIII explains the sorting without any new facts. Regulated utilities hold coherence effortlessly — not because they adapted well, but because the game moved onto their home field: rate cases, commission hearings, and cost allocation are the game they were built for, so a rule change that forces everyone else to learn utility procedure is, for them, no change at all. Hyperscalers with standing permitting teams, political relationships, and routing options — dedicated generation, friendlier jurisdictions — hold coherence at a cost. Greenfield developers built for one game — find land, build fast — face the hardest test, because their entire architecture assumed rules that no longer exist. Speed of adjustment is measurable, and it will show in the public record: the firms that reposition within weeks of EO 62 capture the option value described in Section V, while the firms that wait for legal clarity watch the window close.

Behavioral economics explains the part of this story that cost-benefit analysis cannot: why data centers keep losing local votes even where the tax math favors them. Loss aversion — the finding, from Daniel Kahneman and Amos Tversky’s prospect theory, that people weigh losses roughly twice as heavily as equal gains — runs through every level of the record this paper assembles. Residents at a planning-board hearing weigh feared losses (water, higher bills, a changed town) more heavily than promised gains (jobs, tax revenue), which is why Loudoun-style tax arguments underperform at podiums and why Washington County’s commissioners voted 4–1 against a $57 billion pipeline of statewide investment. Governor Mills’s veto in Maine follows the same logic from the opposite side: one concrete, visible project loss in the town of Jay outweighed diffuse statewide risks. And the incumbents’ silence follows it too — companies now holding scarce permits behave exactly as endowment effects predict, quietly protective of a windfall they never had to argue for.

Loss aversion also operates inside the permit office, and it explains why delays persist even where no one opposes a project. MindCast’s NIBE work measures what it calls the Risk Interpretation Index: the mid-level reviewers who control actual approval timelines systematically overweight approval risk, because an approved project that later fails can damage a career, while a delay carries no penalty at all. The federal-scale demonstration came from the White House Genesis Mission analysis, which found that repairing those mid-level incentives accelerates deployment roughly three times more than adding senior political oversight does. The finding carries directly into this paper’s framework: plank three’s shot clock works not because it hurries anyone, but because it finally gives delay a cost — correcting the one-sided incentive that makes slow the safest thing a reviewer can be.

Loss aversion also explains why the model framework in Section XI is built the way it is. Cost causation works politically, not just economically, because it removes the feared loss instead of promising a compensating gain: a resident guaranteed that their rates cannot rise needs no persuading about tax revenue. Developers who lead with guarantees against losses — rate protection, water commitments, the financial assurances in Baumgartner’s bill — will out-negotiate developers who lead with promised benefits, and states that write loss-protection into statute will site facilities that states relying on benefit-promises cannot. The behavioral read and the game-theory read converge on the same forecast: the fourth cycle rewards the actors who understand which game is now being played, and in the new game the currency is protection, not promises. Section X states what would prove all of this wrong.

X. What Would Falsify the Thesis

A framework earns credibility by naming its own failure modes before critics do. The regulatory repricing thesis has four. Each is described plainly below, with a probability, and the prediction register in Section XII inherits their conditions. Naming them also clarifies what this paper does not claim: it does not claim the AI buildout stops, and it does not claim every moratorium survives.

Failure mode one: courts strike or narrow the model. EO 62 pauses permits statewide by executive order, without new legislation. An order of that shape invites three principal theories: the governor exceeded delegated authority under New York law; queue holders or applicants suffered a compensable deprivation; or the pause imposes an unconstitutional burden on interstate commerce. The dormant Commerce Clause claim is plausible but not the cleanest path. EO 62 appears facially neutral and invokes conventional local interests — reliability, water, air quality, and ratepayer protection — so a challenger would need unusually strong evidence of protectionist design or a burden clearly excessive relative to those benefits. A statutory-authority or administrative-law claim may therefore reach the merits sooner.

One Commerce Clause theory deserves fuller treatment than the standard analysis gives it: extraterritorial effects. Data centers are not local businesses that happen to be large; they are nodes in inherently interstate markets. New York capacity serves cloud customers, financial-services routing, and latency-sensitive workloads across state lines, so a statewide construction pause does not stay inside state borders — it constrains supply in a national compute market and shifts traffic, prices, and siting decisions in other states. National Pork Producers Council v. Ross (2023) cuts both ways on that theory. The decision closed the door on near-automatic extraterritoriality claims: the Court upheld a California law whose practical costs fell mostly on out-of-state producers, holding that in-state regulation does not become unconstitutional merely because its economic effects travel. But the decision left Pike balancing alive — a challenger can still argue that a law’s burden on interstate commerce is clearly excessive relative to its local benefits. The strongest fact pattern for a challenger is accordingly narrow: an out-of-state cloud customer or interconnection applicant showing that New York capacity has no practical substitute for a latency-bound interstate service, and that a pause delivers little incremental local benefit over the cost-causation tariffs that could protect ratepayers without freezing supply. Even that framing runs uphill. Courts rarely strike facially neutral utility and land-use measures under Pike, and New York’s published exit condition weakens any excessive-burden story. The realistic function of the extraterritoriality theory is leverage rather than victory: pleading it raises the legal cost of extending the moratorium past its stated window — which reinforces the register’s first prediction rather than threatening it. A court striking or narrowing the order would not kill the identity-shift thesis, which now stands independently on FERC, Seattle, county, and federal-executive action, but it would end New York’s status as the model and force AIRC-III-1 to resolve early. Probability of a formal legal challenge within six months: 50–60%. Probability a dormant Commerce Clause theory is pleaded: 30–40%. Probability a Commerce Clause theory materially succeeds before the GEIS completes: 15–25%. Probability any challenge significantly narrows the order before the study finishes: 25–35%.

Failure mode two: companies route around permission instead of paying for it. Hyperscalers have countermoves the moratorium wave cannot reach: building their own power generation on site, buying campuses that already have power, using federal land offered under EO 14318, and siting in Texas, where the grid sits outside federal jurisdiction — the same combination of moves Chevron and Microsoft used when they built dedicated generation for an AI campus inside ERCOT. If routing around regulation scales faster than regulation spreads, permission never becomes truly scarce, and this repricing fades into a two-year inconvenience. Probability that routing strategies substantially blunt the repricing within 24 months: 30–40%. Note what this failure mode does not threaten: routing around regulation requires enormous capital, so only the largest firms can do it — which accelerates the sorting of winners from losers rather than preventing it.

Failure mode three: demand falls. The whole framework assumes the hunger for megawatts continues. Two things could relax the constraint from the demand side: AI models becoming efficient enough that computing needs shrink, or an investment correction that cancels queued projects before regulators ever rule on them. History leans against the first — efficiency gains have usually increased total consumption of computing, not decreased it — and 12 gigawatts of pending requests in New York alone suggests demand is outrunning efficiency for now. But the question is genuinely open. Probability that demand-side relief substantially loosens the permission constraint by 2028: 20–30%.

Failure mode four: the numbers favor the buildout. The moratorium wave is real and, so far, small. Wisconsin alone has seven major projects worth more than $57 billion moving forward. In Loudoun County, Virginia, data center taxes fund cuts to residents’ tax rates — the argument the lone dissenting commissioner made against Washington County’s pause. And a large customer with a long-term contract can lower everyone’s rates by spreading the grid’s fixed costs across more usage. Counted against total national construction, the pausing jurisdictions are a small minority. The paper’s predictions concern the trend, not today’s totals — and a reader who weighs today’s totals more heavily concludes the repricing is two or three years early, not wrong. The register exists to settle exactly that kind of dispute.

The survey changes the paper’s posture without changing its content. Regulatory repricing is the most likely outcome, not a certainty. The thesis’s strongest challenge is companies routing around regulation, not courts. The thesis’s weakest assumption is that demand stays high, not that politicians stay willing. Analysis this close to live policy also owes readers a position, and Section XI states MindCast’s.

XI. The MindCast Position and a Model Framework

Analysis that stops at description ducks the question every reader of this paper will ask: are New York and Washington doing the right thing? MindCast’s answer follows directly from the paper’s own findings, and it separates the goal from the tool.

Start with where MindCast sides with the industry, because on the economics the buildout’s case is stronger than the moratorium wave’s rhetoric admits — and three facts anchor it. First, some of the largest operators are already internalizing substantial portions of the infrastructure burden. The signature moves of the past two years include dedicated generation deals, restarted nuclear plants, newly funded geothermal, and long-term contracts that can add supply to the system rather than merely consume existing capacity. Under this paper’s own test, that is constraint-removal capital — spending that increases the number of viable deployment paths — and it is being deployed voluntarily, at private expense, by the very firms the moratoriums target. Second, the 12-gigawatt queue indicts both grid-development speed and queue governance. Some requests may be speculative, but a backlog that deep still shows that the interconnection system operates at a pace built for a slower load-growth era. Pausing demand because supply processing is slow gets the causation backwards, and no state ever fixed a slow permit office by turning away the applicants offering to fund its expansion. Third, the benefits are concrete, not promised. Where the buildout is mature, data center tax revenue funds cuts to residents’ tax rates; a large customer on a long-term contract spreads the grid’s fixed costs across more usage, which can lower everyone else’s bills; and the national stake is real — the AI race is being run whether or not any given state participates, and the compute will be built somewhere.

The industry’s position is not inferred — it sits in Washington’s official hearing record, where the Data Center Coalition testified that the industry is “fully committed to paying full costs” and to protecting other customers from stranded assets (Senate bill report, SB 6171). The position compresses into one sentence: a willing payer facing a closed path is a policy failure, not a corporate one. The right response to a customer offering to pay the full cost of new infrastructure is to take the money and build the grid — which leaves every other ratepayer with a bigger, better-funded system. Both states get graded against that standard below.

The goal both states are reaching for is correct, and it now has a name both parties use: cost causation. A facility that requires new generation, transmission, or water infrastructure should pay the full incremental cost of it — not ratepayers, not taxpayers. Hochul’s grid fund says this. Baumgartner’s bill says this. The principle is economically sound and MindCast endorses it without reservation.

The moratorium, however, is the wrong tool for that goal, and Section V already showed why. A pause hands windfalls to incumbents, invites litigation, pushes demand to Virginia and Texas rather than reducing it, and produces nothing a ratepayer can spend. Pricing achieves the same protection without the distortions. In one sentence: charge them, don’t pause them.

New York therefore gets a conditional grade: right goal, second-best tool, defensible only because the order published its own exit. If the state lifts the moratorium under a payment framework within its stated year, EO 62 will have worked as a forcing device. If the moratorium extends past July 2027, it stops being ratepayer protection and becomes a state-enforced moat for the companies that already hold permits. AIRC-III-1 in the register below is therefore not just a forecast — it is the test New York set for itself.

Washington gets a harder grade: the state stalled at the last gate, and the stall produced a worse outcome than either passing or killing House Bill 2515 cleanly would have. The House passed the bill 51–41; the Senate policy committee advanced it with amendments; the Senate fiscal committee then let it die without a vote. A near-enacted framework left a vacuum anyway, Seattle filled it, and the state now has a city-by-city patchwork with thresholds no statehouse would have chosen — bad for developers who face inconsistent rules, bad for cities negotiating alone against the world’s largest companies, and bad for ratepayers outside Seattle, who get no framework at all. MindCast’s position: Olympia should pass a comprehensive statewide framework in the 2027 session, built on cost causation, superseding the patchwork — one set of rules serving both Seattle’s grid concerns and Eastern Washington’s recruitment ambitions. Baumgartner’s payment and water-reuse principles are the right economics; whether they belong in federal or state law remains genuinely open, because one national rule fits hydro-rich Washington and gas-dependent states differently.

Criticism obligates a proposal, so MindCast offers one: a model state framework with five planks, each built to prevent one of the failure patterns this paper documents.

Plank one: charge, don’t pause. Applications keep processing under plank one; only the price changes. Any facility above the state threshold pays the full incremental cost of the generation, transmission, and grid upgrades it causes. The facility posts financial assurances before connecting. And it takes service under a dedicated large-load rate class, so its costs never blend into residential bills. Plank one combines Hochul’s fund, Baumgartner’s bill, and Seattle’s rate class into a tariff instead of a freeze — protecting ratepayers without minting incumbent windfalls or inviting lawsuits.

Plank two: one threshold per state, with local consent guaranteed by law. Three governments defining the same facility at three sizes is pure waste. The state should set a single definition and preempt local thresholds — and pay for that preemption by writing local leverage into statute: no siting without host-community approval, plus defined community-benefit, wage, and local-hire standards. Cities trade improvised moratorium power for guaranteed bargaining power. Plank two is the deal that prevents the Washington patchwork. Quantified support comes from MindCast’s own Washington analysis: the state’s seven-layer approval stack — federal power marketing, FERC, state agencies, counties, cities, public utility districts, and tribal governments — costs it an estimated $30 to $50 billion in investment lost to Texas, Virginia, and Oregon, and the highest-leverage fix identified was consolidation into a single decision point, because consolidating authority beats optimizing each institution separately.

Plank three: a shot clock, never an open-ended pause. A state that genuinely needs a study period should authorize it by legislation with a hard deadline: permits resume automatically when the clock expires unless new rules have been enacted. A deadline written into statute does what a deadline announced at a press conference cannot — survive a lawsuit and resist drift. Plank three is the direct answer to New York’s largest risk.

Plank four: the incumbent windfall check. Section V showed that every grandfathering clause enriches the companies it exempts. The fix: exempt existing projects from the pause but not from the pricing — cost-causation charges reach incumbents at contract renewal or expansion. Ratepayer protection should not double as a moat.

Plank five: water and queue hygiene. In aquifer-dependent regions, require recycled water and offer a reuse tax credit — Baumgartner’s mechanism, generalizable to any state. And require refundable deposits on interconnection requests to flush speculative applications from the queue: New York’s 12-gigawatt backlog almost certainly contains phantom projects that will never build, and no state can plan a grid around requests that exist only on paper.

The one-line version: a state that adopts all five planks never needs a moratorium, and a state already inside one has its exit ramp. New York needs planks three and four to avoid becoming the thing it claims to oppose. Washington needs planks one and two in the 2027 bill to dissolve the patchwork it accidentally created.

One test evaluates every plank — and every future rule any state proposes — at once, and it comes from the Power Stack series: does the rule increase or decrease the number of places a project can viably be built? MindCast calls that count the Geodesic Availability Ratio (GAR) — in plain terms, how many viable deployment paths exist across the system. A moratorium closes paths and fails the test. A cost-causation tariff prices paths without closing any and passes. The test resolves the apparent conflict between protecting ratepayers and staying competitive: a rule that allocates costs correctly while keeping paths open does both, and a rule that closes paths does neither for long — it just moves the projects, the jobs, and eventually the tax base to a jurisdiction that kept its paths open.

Stating a position and proposing a remedy sharpens rather than compromises the foresight record, because both concern instrument design, not partisan outcome — and the register that follows scores them. If states converge on cost-causation frameworks, the grid-fund predictions resolve true; if the moratorium wave spreads instead, the propagation predictions resolve true and the framework above becomes each new state’s exit ramp. Section XII lists the entries.

XII. Prediction Register — AIRC-III

The register uses the AI Repricing Cycle (AIRC) namespace, entries AIRC-III-1 through AIRC-III-14. MindCast publishes forecasts the way an exchange lists contracts: each entry has a deadline, a confidence band, and a condition that proves it wrong. The federal intervention changes the register’s hierarchy. FERC has already opened the large-load proceeding, so the question is no longer whether federal intervention begins; the question is what remedy it produces and whether partial federalization holds. Several entries share causal drivers: AIRC-III-1 and AIRC-III-3 depend on cost-causation convergence; AIRC-III-2 and AIRC-III-7 depend on political propagation; AIRC-III-4 and AIRC-III-14 arise from the same FERC proceeding; AIRC-III-9 and AIRC-III-13 test distinct judicial and legislative routes. The fourteen entries should

Every entry resolves from public records. The New York DPS and governor’s office decide AIRC-III-1 and AIRC-III-5; state statutes and commission dockets decide AIRC-III-2, AIRC-III-3, and AIRC-III-8; FERC Docket No. EL26-69-000 decides AIRC-III-4; published research decides AIRC-III-6; municipal records decide AIRC-III-7; court dockets decide AIRC-III-9; transaction disclosures decide AIRC-III-10; corporate announcements and filings decide AIRC-III-11 and AIRC-III-12; the Congressional Record, White House, agency notices, or federal pleadings decide AIRC-III-13; and the combined federal, state, and judicial record decides AIRC-III-14. The register remains live through July 2028, with resolutions published as they land.

XIII. Conclusion

EO 62’s own text makes the paper’s argument better than any analyst can. The order defines a data center by its power consumption, not by what it computes. When the regulator classifies an industry by megawatts, the market should value it by megawatts. New York did not invent this cycle — the queue backlog, the ratepayer politics, and the grid limits built it over two years. The executive order simply made the cycle visible, and visibility is what triggers repricing.

The end state also has a name. A companion to this paper — the AIRC-III Cognitive Digital Twin Library — models the fourteen actors who each control a portion of the authorization pathway: governors, environmental agencies, system operators, hyperscalers, developers, utilities, competing states, labor, ratepayer coalitions, investors, litigants, and analysts. No actor controls the whole pathway; several can delay it. Their interaction does not converge on a ban and does not converge on an unconditioned buildout. It converges on conditional acceleration: FERC standardizes parts of the transmission pathway, states and cities retain site-specific leverage, and large-scale AI infrastructure proceeds where developers accept differentiated tariffs, prepay grid costs, secure local consent, or route through private and federal capacity — while stalling where those terms remain unavailable or unpriced. Conditional acceleration is this paper’s modal forecast, the register is its scoreboard, and the model framework in Section XI is its recommended legal form.

Artificial intelligence has not become less valuable. Value has moved — into infrastructure economics, where permission compounds like capital and permitting expertise is worth what patents used to be. The first generation of AI winners accumulated GPUs. The second generation accumulated inference demand. The third generation will accumulate megawatts — and the permission to use them.

Appendix A — Acronym Reference

Appendix C — MindCast AI Works Cited

The AI Repricing Cycle

• AI Repricing Cycle 2026 — Installment I — Establishes the repricing-cycle framework and documents the capital-scarcity phase that this paper extends to authorization.

• Escape from the AI Repricing Cycle — Installment II — Prices the inference-cost phase and sets the July 29 Microsoft earnings checkpoint that Section V uses as a live test date.

• The Inference Economy — How the TPU Bifurcation Repriced the AI Compute Stack — Documents the compute-stack repricing that made power, rather than chips, the next binding constraint.

Energy, Infrastructure, and the Grid

• The Power Stack: How Energy Infrastructure Became the New AI Battleground — Forecast the trigger conditions for government intervention that EO 62 satisfied; Section I scores the resolution.

• The AI Infrastructure Energy Opportunity Landscape — Supplies the constraint-removal versus choke-point distinction and the Geodesic Availability Ratio test applied in Sections V and XI, and forecast the state-by-state permitting divergence documented in Sections VI and VII.

• The AI Infrastructure Energy Antitrust Landscape — Maps the enforcement exposure that Section V’s self-reinforcing loop predicts moratorium windfalls will attract.

• Chevron–Microsoft Project Kilby and the Firm-Power Forecast — Documents jurisdiction selection as a regulatory routing layer; the ERCOT precedent behind Sections IV, VII, IX, and X.

• The Federal-State AI Infrastructure Collision — Forecast partial federalization, the coordination tax, and New York’s conditioning posture; Section VII tracks each against the FERC §206 proceeding.

• The DOJ–FTC Gas-Price Letter to State Attorneys General — Establishes the federal energy-market intervention posture that frames the enforcement trajectory in Sections V and X.

Method Frameworks

• Predictive Game Theory Meets the Era of AI — Operationalizing Fudenberg’s Research Agenda with Cognitive Digital Twins — Supplies the game-replacement framework and Adaptive Coherence Equilibrium concept applied to the actor field in Section IX.

• The MindCast AI Dynamic Predictive Game Theory Collection — Collects the foundational game-theory pieces behind the Section IX analysis in one visual edition.

• Synthesis in National Innovation Behavioral Economics and Strategic Behavioral Coordination — Quantifies institutional throughput: the five-to-one temporal mismatch in Section VII, the reviewer loss-aversion finding in Section IX, and the Washington seven-layer analysis in Section XI.

• White House Genesis Mission x NIBE — Demonstrates that correcting mid-level reviewer incentives outperforms senior oversight, the finding that gives plank three its mechanism.

Law and Governance

• The Commerce Clause as America’s AI Advantage — Establishes the constitutional path for federal AI infrastructure coherence and the absorption model behind Section VII and prediction AIRC-III-13.

• Susquehanna Is Building the Institutional Market the CFTC Does Not Require — but Still Needs — Documents the parallel legitimacy game: building the record a regulator politically needs before the law requires it, the same structure as New York’s grid-fund exit.

Academic Foundation

• Kahneman, D., & Tversky, A. (1979). “Prospect Theory: An Analysis of Decision under Risk.” Econometrica47(2): 263–291 — The loss-aversion mechanics applied at the community, gubernatorial, and reviewer levels in Section IX.

I. Executive Summary

Thesis: Susquehanna’s World Cup hedging offer builds a record the Commodity Futures Trading Commission (CFTC)does not legally require but politically needs.

Before 2000, the CFTC applied an economic-purpose test when determining whether futures contracts satisfied the statutory public-interest requirement. The 2000 Commodity Futures Modernization Act removed that test, and Dodd-Frank never restored it. A sports event contract can now clear the Commission’s public-interest inquiry without demonstrating any hedging function.

Susquehanna is not repairing the doctrine behind CFTC jurisdiction, because the Commission already claims exclusive jurisdiction on statutory grounds and does not base that claim on economic purpose. Susquehanna is supplying the concrete institutional example that makes the Commission’s permissive treatment easier to defend before courts, Congress, and a skeptical public.

One distinction runs through every section. An offer of capacity is not proof of demand. Susquehanna has built the offer side of a possible institutional market, and until a named counterparty executes a disclosed hedge, the demand side stays invisible and the facility works as legitimacy support rather than as evidence of a real market.

The forecast in brief, five falsifiable calls scored in a pre-resolution robustness audit:

Section II states the calls in full. The analysis that follows explains each one.

II. The Five Calls

MindCast lists forecasts the way an exchange lists contracts, each with a dated window and a falsifier. Three tools generate and score them: Dynamic Predictive Game Theory for the sequencing and best-response structure of the regulatory game, Behavioral Economics for the retail-versus-institutional composition of the market, and the MindCast AI Proprietary Cognitive Digital Twin Foresight Simulation (MP CDT FS) engine for the pre-resolution robustness audit that set the bands. The five gates below are tightly observable.

Two pairs share causal ground and should not be read as independent confirmations. EP-1 and EP-4 draw on the same comment-and-litigation record. EP-2 and EP-3 both turn on corporate disclosure behavior, and can move in opposite directions as that behavior shifts.

EP-1, the commercial-use record. At least one CFTC comment on Regulation Identifier Number (RIN)3038-AF65, or one court filing in a prediction-market federal-preemption or Rule 40.11 docket (the Fourth Circuit, Massachusetts Supreme Judicial Court, or Kaiserman v. Kalshi among them), cites Susquehanna’s hedging capacity as favorable public-interest evidence under § 40.11(a)(5). Falsifier: no qualifying citation by the end of Q1 2027. Window: through Q1 2027. Band: 68–77%.

EP-2, capacity versus demand. No named counterparty publicly discloses a material World Cup hedge, at least $1 million notional, through the facility before kickoff of the July 19 final. Nondisclosure proves only the absence of public validation, not the absence of private demand. Falsifier: a named sponsor, broadcaster, or brand discloses a $1 million-plus hedge through the facility before kickoff. Window: through kickoff of the July 19 final. Band: 78–86%.

EP-3, disclosed affiliated routing. Public disclosure shows the facility’s flow executing on Rothera or another Susquehanna-affiliated venue. The gate turns on disclosure, not underlying routing: affiliated routing may well occur yet never surface in a filing, which pulls the band down. Falsifier: disclosed routing to an unaffiliated venue, or no routing disclosed. Window: through Q4 2026. Band: 38–48%.

EP-4, the litigation argument. At least one state or private litigant argues that limited institutional use cannot outweigh predominantly recreational retail participation. The gate registers the argument’s appearance, not a court’s holding. Falsifier: no such argument appears in comments or filings by the end of Q1 2027. Window: through Q1 2027. Band: 69–78%.

EP-5, aggregate adoption without disclosure. Susquehanna, Rothera, Robinhood, or an industry participant characterizes the facility as evidence of institutional adoption without naming a counterparty, size, price, and venue together. Falsifier: every adoption claim supplies all four elements, or none appears. Window: through Q4 2026. Band: 72–82%.

Mechanism and pricing gates register in the separate quarterfinal Validation Report; the volume-cliff and dual-citation gates (P2, P4) stay live in the foundation register; the integrity and surveillance gates stay in the internal register.

III. The Test the CFTC Rejected

Read on its own terms, the June 10 proposal opens by dismantling an assumption most commentary still makes. Before 2000, the CFTC applied an economic-purpose test when deciding whether futures contracts met the statutory public-interest requirement. The 2000 Act removed that test, Dodd-Frank declined to revive it, and nothing in the proposed Rule 40.11 brings it back.

A sports contract can now clear the public-interest inquiry without any business ever hedging with it.

What the Commission kept is a weighing test, not a gate. The favorable side lists meaningful risk transfer, price discovery, commercially useful information, and responsible innovation. The unfavorable side lists manipulation risk, weak settlement integrity, and insider exposure.

The Commission then qualifies its own favorable factors in ways that decide this case. No single factor controls. Direct hedging use carries, in the proposal’s phrase, limited importance. A sports-contract price can serve an economic purpose even when no commercial party trades it. And properly designed professional and collegiate sports contracts already look unlikely to fail the public interest wherever an exchange keeps adequate integrity controls.

The proposal separates four ways a contract can carry economic value, and only one of them requires a hedger. Direct hedging use puts a real position on to offset a real exposure. Price discovery lets the market’s odds inform a decision. Information aggregation pools scattered knowledge into a single number. Commercial use of that number lets a business act on it without ever trading.

A hotel near a host stadium shows the difference. The hotel need not buy a single World Cup contract to use its price. A rising probability that the home team advances tells the hotel to raise room rates, add staff, and stock inventory for another week of demand. The contract’s price does economic work while the hotel stays out of the market entirely.

The distinction is why Susquehanna’s facility helps the Commission without being necessary to it. Institutional hedging strengthens the direct-use factor, the narrowest of the four, while the other three support the contracts on their own.

Permissive treatment of aggregate-outcome sports contracts therefore sits close to a baseline, not a prize an applicant must win with a hedging record.

The gap between what the rule requires and what the rule rewards is the entire opening. The Commission needs no commercial hedging to permit these contracts, yet it reaches for hedging, price discovery, and commercial usefulness again and again to justify permitting them.

A regulator still has to defend a permissive rule, before a Fourth Circuit panel, a state supreme court, a hostile congressional committee, and a wary public. Abstractions defend it less well than a concrete institutional example does. The favorable factors are legally non-controlling and politically load-bearing at the same time, and the difference between those two is the whole game.

The reading comes straight from the proposal’s text rather than from any inference: confidence 95–99%.

IV. Susquehanna Builds the Offer Side

Susquehanna’s announcement borrows the proposal’s commercial-risk vocabulary almost word for word, and the borrowing is the point.

The firm offered up to $500 million in capacity to help institutions hedge World Cup outcomes, naming exactly the clients the rule would smile on: sponsors, broadcasters, hospitality operators, and consumer brands whose promotions and rebates ride on which team advances. Line the announcement up against the favorable factors and the language matches, risk transfer, commercial utility, a legitimate hedging use.

An offer of capacity is not a market, though, and the distinction governs everything that follows. Susquehanna has built the offer side and nothing more. No public evidence names a deployed dollar, an executed hedge, a counterparty, a size, a price, or a venue.

What the facility can do is turn a regulatory abstraction into a citable commitment, a commercial fact pattern ready for the record the moment the permissive rule is challenged.

Timing carries the same double meaning. Capital announced inside the comment window, dressed in the language of the public-interest factors, reads as a fact pattern staged for the record, even as the approaching final gives Susquehanna an ordinary commercial reason to move now. The two motives coexist rather than compete: confidence the offer materially serves record-construction alongside commercial origination, 62–72%.

Subscribe now

Contact mcai@mindcast-ai.com to partner with us on Predictive Game Theory AI in Law and Behavioral Economics. Our verticals include simulating complex litigation, innovation economics and geopolitical risks.

To deep dive on MindCast works upload the URL of this publication into any LLM (preferably Google AI mode) and prompt ‘reconstruct MindCast framework with three degrees of cited sub links.’ See Appendix A and Live-Fire Game Theory Simulators, Runtime Predictive Infrastructure.

We stress-tested our AI system for this publication by simulating the Super Bowl and the World Cup.

V. How the Hedge Would Actually Work

A concrete example turns the abstraction into mechanics. Picture a consumer brand that promises customers a rebate if the national team it sponsors wins the World Cup. Each round the team survives raises the expected rebate the brand will owe.

The brand buys contracts that pay if the team reaches the triggering stage. A favorable run then moves two cash flows at once: the rebate liability climbs, and the contract payout climbs with it, so one offsets part of the other.

The offset is partial by design. The contract hedges the triggering event, not the full economic loss, and the size of the position sets how much of the liability it covers. A brand that buys too little stays exposed, and a brand that buys too much overshoots into a speculative position.

Basis risk lives in the gap between the contract’s settlement terms and the actual liability. A contract that pays on a semifinal berth does not match a rebate that triggers on the title, and closing that mismatch is the hedger’s job to size.

The choice between venues is a real tradeoff, not a foregone conclusion. An exchange offers standardized pricing and divisible exposure, so a treasurer can buy a precise notional in a liquid market. Insurance offers tailored coverage written closer to the actual loss, at the cost of liquidity and speed.

Susquehanna sits on the supply side of the exchange route. The firm supplies liquidity against the institutional order while the exchange lists and clears the position, which is why the announcement reads as capacity rather than as a placed trade.

The phrase itself measures room to transact, not money at work. “Up to $500 million” describes available transaction capacity, not executed volume and not reserved cash, so the figure sets a ceiling on what the facility could absorb rather than a floor under what it has done.

VI. Capacity Is Not Demand

Dynamic Predictive Game Theory treats the offer as a move in a repeated regulatory game, and a strategic claim in that game settles only when two conditions hold at once. The players’ behavior has to converge, the game-theoretic condition, and enough information has to exist to confirm the claim, the informational condition. MindCast’s framework names the joint requirement the Dual-Equilibrium Termination Architecture (DETA), taking the first from Nash and the second from Stigler.

Susquehanna’s offer meets neither cleanly, and the shortfall is the forecast.

Take the behavioral condition first. A public offer of execution capacity is a real, observable, reputationally costly signal, and a firm of Susquehanna’s size needs no segregated vault to make it credible. The announcement appears to represent fungible execution capacity rather than segregated capital, though the public record does not disclose the internal allocation.

The signal’s defensive value holds even if no hedge executes, because a citable commitment lends narrative support to the venues Susquehanna co-owns regardless of use. One firm moving, though, is not convergence. The Commission set the game by proposing the rule; Susquehanna merely moved first among market participants.

The informational condition stays open, and it does the real work. Enough evidence to confirm genuine demand does not yet exist: no disclosed counterparty, executed hedge, size, price, or venue. Capacity creates the test; only execution produces the evidence that passes it.

The offer moves through four states, and each one settles a different question. First, the offer is announced, which is where the record sits today. Second, a counterparty shows interest, which validates curiosity and nothing more. Third, a transaction executes, which validates private demand. Fourth, the transaction becomes publicly observable, which is the only state that validates the public legitimacy narrative.

The distance between the third state and the fourth is the whole forecast. A hedge can execute privately and never surface, so demand can be real while the public record stays empty.

The open layer is demand, not defensive value. The offer already works as legitimacy support, and only its weight as proof of genuine commercial use waits on execution.

Counterparty confidentiality, authorization lead times, and the short tournament window make public validation less likely than private execution. A pre-resolution MP CDT FS audit therefore raises the confidence that no named counterparty discloses a material hedge before kickoff of the July 19 final to 78–86%, with silence proving only the absence of public validation, not the absence of private demand.

VII. The Insurance Counterfactual

The clearest real case of sports-outcome hedging never touched a prediction market, and it sets the bar Susquehanna has to clear.

Club Atlético Osasuna paid €1.2 million for an insurance policy that would have returned €6 million on relegation, a bespoke bilateral contract written for a party with real exposure. Insurance and tailored bilateral deals already hedge sports outcomes for the entities that carry the risk, so institutional demand for this protection is not in question.

Whether an exchange-traded event contract is the efficient way to supply it is very much in question, and both sides of the comparison carry weight.

A bespoke policy can be illiquid, costly to structure, and difficult to transfer. An exchange-traded contract can offer standardized terms, transparent pricing, divisible exposure, and centralized clearing that reduces bilateral counterparty exposure. Whether it offers superior capital efficiency depends on the contract’s collateral and margin requirements.

The countervailing fact is behavioral, and equally real. The entities with genuine exposure already hedge through insurance and bilateral deals, while retail speculation appears to supply a substantial share of sports-contract activity, the market-composition signal Behavioral Economics tracks. The open question is whether the efficiency gain mainly serves hedgers or mainly lowers the friction on speculation.

Neither the proposed rule nor existing doctrine forces that showing as a legal matter, yet the answer sets how persuasive the institutional-use story becomes. A litigant will likely press the comparison, arguing that institutional hedgers already have insurance and bespoke markets: confidence that argument appears in comments or filings by Q1 2027, 68–78%.

VIII. The Legitimacy Game

Susquehanna’s offer plays in a legitimacy game, not a jurisdiction game, and separating the two dissolves the strongest temptation in the analysis. Two paths run side by side, and the offer touches only one.

The Legal Path

Economic use does not move classification. A sponsor’s hedge does not turn a touchdown-total contract into a derivative, and it does not push a sports contract into exclusive federal jurisdiction.

Contract structure, statutory definitions, registration status, and the reach of federal preemption decide those questions, and a $500 million offer touches none of them. The legal path runs on statute and precedent, and no commercial fact pattern reroutes it.

The Legitimacy Path

What the offer touches is defensibility, the record the Commission leans on when courts, Congress, and the public ask why sports contracts should list freely.

Three players move together, and the offer resequences all of them. The Commission, defending a rule now that Loper Bright has ended mandatory Chevron deference, gains a concrete example where it once had only abstraction. A well-developed administrative record can still support the rule as reasoned agency decision-making under arbitrary-and-capricious review, and courts may credit that reasoning under Skidmore even without any institution publicly disclosing a trade, so the fact pattern strengthens the political defense rather than supplying a missing legal element.

The platforms, whose growth products draw heavy retail speculation, gain an institutional-use story that softens the gambling-site framing. Susquehanna, first to move, converts capacity into standing in the record before any rival can.

The system keeps settling into the same split, contests to the states and consequences to the CFTC. The offer is a legitimacy bid on the federal side of that line, not a lever that moves the line.

The Vertical-Integration Question

Susquehanna’s own shape raises the sharper question. Firms like Susquehanna, Jane Street, and Citadel have priced these events for years, but privately, for their own books, making no public claim, what MindCast’s field map calls proprietary probability engines.

Susquehanna now occupies several layers of a public exchange at once: market maker on Kalshi, co-owner of Rothera through its joint venture with Robinhood, and potential originator of the very institutional flow the facility promises. The same offer that dresses retail speculation in institutional-risk vocabulary also lends legitimacy support to the venues Susquehanna co-owns.

The footprint creates a real conflict-management question about affiliated routing and captive liquidity. The footprint does not establish self-dealing, and the honest reading says so: confidence the affiliated-venue question surfaces in commentary or filings by Q4 2026, 55–65%.

IX. What the Paper Means for Each Stakeholder

Susquehanna built one object that every actor can translate into its own language. The Commission sees commercial utility it can cite. The platforms see institutional maturation. The states see a thin institutional layer resting on a much larger base of retail speculation. The counterparties see one more risk-transfer tool to price against insurance. And Susquehanna sees liquidity, influence over the record, and optionality across the venues it touches.

Susquehanna built one object that every actor can translate into its own language. The Commission sees commercial utility it can cite. The platforms see institutional maturation. The states see a thin institutional layer resting on a much larger base of retail speculation. The counterparties see one more risk-transfer tool to price against insurance. And Susquehanna sees liquidity, influence over the record, and optionality across the venues it touches.

Each reading is internally consistent, which is what makes the offer effective. One disclosed, material, executed hedge would convert the offer from a fact pattern into shared evidence. Continued silence leaves each camp holding its own interpretation and the central claim unproven.

X. Conclusion

Susquehanna cannot manufacture jurisdiction by offering hedging capacity, and it has not tried to. The Commission already claims exclusive jurisdiction over these contracts on statutory grounds, so institutional hedging does not change the statutory classification analysis or independently resolve who governs them.

What the offer changes is the quality of the Commission’s public explanation. A regulator defending free-listed sports contracts argues from firmer ground with a concrete institutional example than with favorable-factor abstractions alone. The favorable factors are legally non-controlling and politically load-bearing at once, and Susquehanna has offered institutional support for that load.

Value here resolves on execution, not announcement. Capacity creates the test the institutional-use story must pass; executed, disclosed, material hedging would close that gap, though the Commission can still defend sports contracts on other public-interest factors.

Until execution arrives, the record holds an observable offer rather than an observable market, and the distance between the two is where the legitimacy contest plays out.

Appendix A: MindCast AI Works

1. The CFTC NPRM Is a Litigation Brief. The reading of the June 10 rulemaking behind Section III: the rejected economic-purpose test, the non-controlling favorable factors, and the treatment of properly designed sports contracts.

2. The Full Arc of Prediction Markets. The field map behind Section VIII, which classifies firms like Susquehanna as proprietary probability engines and frames the affiliated-routing conflict without overclaiming self-dealing.

3. The Dual Nash-Stigler Equilibrium Architecture. The closure logic behind Section VI: a claim settles only when behavioral convergence and informational sufficiency hold together.

4. How MindCast Evolves the Structural Gaps in Classical Nash Game Theory. The game-theoretic frame for the sequencing advantage the offer exploits.

5. The CFTC’s Undefined Category. The companion on the classification-jurisdiction axis, scoring the July 7 New York ruling.

6. The World Cup Is the First Mass-Market Stress Test for Prediction Markets. The foundation piece supplying the four-engine framework and the P2 and P4 gates this register inherits.

7. Defining “Gaming” Under the Commodity Exchange Act: A Rule 40.11 Framework. MindCast’s April 17 comment on RIN 3038-AF65.

8. MindCast AI Comment on the Prediction Markets NPRM. MindCast’s June 10 comment on the same docket.

Appendix B: Primary and External Sources

9. Susquehanna Announces World Cup Hedging Capacity of Up to $500 Million. BusinessWire, July 8, 2026. Establishes the offer as up to $500 million in capacity, the offer side, not deployed capital.

10. CFTC Notice of Proposed Rulemaking, “Prediction Markets; Public Interest Determinations,” RIN 3038-AF65. Published June 10, 2026; comments due July 27, 2026; Press Release 9249-26.

11. Osasuna, Clarification on the Contracting of Relegation Coverage. The €1.2M/€6M relegation insurance policy behind Section VII.

12. Robinhood–Susquehanna Prediction Markets Joint Venture. The ownership behind Rothera and the affiliated-routing question in Section VIII.

13. Bloomberg via Yahoo Finance, “Susquehanna Puts Up $500 Million for ‘Hedging’ World Cup Games” (Bernard Goyder), July 8, 2026. Ric Best’s named risk categories and the Osasuna detail.

Executive Summary

No company can stop a stock drop. Markets incorporate new information continuously, discount rates move, sectors rotate — and even exceptionally well-managed firms draw down when the whole AI trade reprices, as June 2026 proved across every layer of the ecosystem. Escape from the AI Repricing Cycle therefore cannot mean immunity. Escape means controlling one variable: registration lag — the time between an internal change in AI economics and its credible external registration.

AI Repricing Cycle 2026: Microsoft, Nvidia, Oracle, Meta and the Validation Tests Every AI Layer Now Faces — hereafter AI Repricing Cycle — diagnosed a recurring validation regime with published crossing dates, a propagation loop connecting six ecosystem layers, and a litigation pattern that punishes narrative gaps rather than cash burn. Doctrine follows diagnosis here: firms cannot escape sector repricing, but they can reduce preventable repricing by shrinking the lag between what the organization knows and what markets expect. One proportionality governs the exposure:

Preventable Repricing Exposure ∝ Registration Lag × Expectation Sensitivity × Representation Surface

Registration lag decomposes into four sequential delays — detection, escalation, adaptation, and communication — and four levers close them one-to-one, while two more address cross-system vulnerabilities. Lag measures how stale the market’s picture is. Expectation sensitivity measures how violently the market moves when the picture updates. Representation surface measures how many statements exist for the update to contradict. Four levers each close a named delay, two address cross-system vulnerabilities, one institutional capability owns the system — and the three-term structure makes the framework’s hardest tradeoff analytically visible — some levers shrink lag while enlarging surface, and managing that exchange is the discipline.

One simulation finding governs everything the levers measure: the most dangerous signal is not missing data — it is accurate data answering the wrong question. Backlog, orders, usage, and capital deployed can all remain accurate while profitable utilization, financing durability, and returns deteriorate underneath them.

Five primary predictions register with this paper. AIRC stands for AI Repricing Cycle, the series’ running registry; Section IX carries falsifiers and an extended conditional tier.

• 🔎 Microsoft (MSFT) validation expansion — 65–75%. July 29 earnings materials materially expand at least two AI-economics disclosure categories.

• ⚖️ Oracle (ORCL) structure registration — 70–80%. Disclosures, financing materials, or the Barrows record distinguish who owns and finances Oracle’s AI infrastructure by December 31, 2026.

• 📊 Gap-lag correlation — 60–70%. The June 2026 decomposition, run under a protocol frozen at publication, will show idiosyncratic drawdown magnitudes rank-correlating positively with ex-ante disclosure-gap scores across the eight-firm cohort.

• ♟️ First-mover reframe — 55–65%. At least one hyperscaler (MSFT, GOOGL, AMZN, META, ORCL) reframes capital guidance as milestone-gated by July 2027, without a surrender-signal drawdown.

• 🏛️ Capital oversight formalization — 55–65%. At least two cohort companies formalize board- or executive-level AI capital oversight by end-2027.

Live tests arrive within weeks, and the calendar chooses the order: Microsoft reports July 29, Barrows v. Oracle briefing runs through December, and Amazon’s free-cash-flow crossing is in progress now.

I. What Escape Means — and What It Cannot Mean

June’s selloff had two components, and the distinction defines everything that follows. Sector beta hit everyone: capex anxiety, rate fears, and rotation repriced the entire AI complex regardless of individual merit. Idiosyncratic repricing hit selectively: Oracle’s (ORCL) 27 percent June drawdown, Microsoft’s (MSFT) alleged demand-capacity gap that supplied plaintiffs a proposed class period, Meta’s (META) share-sale scare — each traced to something the firm represented, structured, or failed to synchronize. Governance touches only the second component. Any framework promising more oversells; any firm demanding more misunderstands the market it trades in.

Preventable repricing has a working definition: the portion of a firm’s drawdown that exceeds what market and sector movement alone would predict, where the excess traces to registration lag — internal changes in AI economics that reached the market late, incoherently, or through a plaintiff’s complaint. June supplies the natural experiment, measured in Section V. The documented record generates the preliminary hypothesis: Oracle and Microsoft will carry the largest idiosyncratic drawdown magnitudes (a financing and lease structure plaintiffs allege the market did not fully understand, and an alleged demand-capacity gap — both registration questions measured in months); Meta’s round-trip should net near zero (a gap opened and closed within days); and Alphabet (GOOGL) and Amazon (AMZN) should land mostly on the sector factors — their lags remain open but unpriced, which is exactly what makes them the registry’s next-defendant candidates. Section V’s model, not the narrative, renders the verdict.

Escape runs on two dimensions, not one. Prevention shrinks the idiosyncratic drawdown; resilience shortens it — the recovery capacity MindCast’s Agent Governance Equilibrium framework (hereafter AGE) measures, with the mechanics in the Foundations appendix. Meta ran June’s proof case: a 7 percent drawdown escalated to a strategic answer within days — the AI-cloud reframe — and the round-trip netted near zero. High resilience is registration-lag control operating in reverse: the firm re-registered its economics faster than the market could harden the gap into a class period.

Escape, precisely stated: a firm exits the cycle’s preventable component when its validation appointments arrive as confirmations rather than surprises — and when the surprises that do land get escalated, owned, and re-registered faster than markets can reprice them.

II. The Six Levers of Registration-Lag Control

Six levers convert the doctrine into operating practice. Each closes one of the four delays named above — or one of the two system-level failures beyond them — and each carries a principal risk the firm accepts by pulling it. The map assigns the pairings; the tour that follows walks each lever’s mechanism, its evidence from June, and its cost.

Theory explains why the levers work rather than merely asserting that they do. MindCast’s Chicago School Accelerated — The Integrated, Modernized Framework of Chicago Law and Behavioral Economics (hereafter Chicago School Accelerated) distinguishes coordination costs from transaction costs (Coase), shows that rational actors drift toward opacity when coordination architecture weakens (Becker), and demonstrates that legal feedback arrives too slowly to correct the drift (Posner). The levers map onto that chain: Levers 1, 5, and 6 build coordination architecture so internal and external expectations can converge; Levers 2 and 3 shift the payoff gradient so honest registration outperforms narrative inflation; Lever 4 and the Section III capability substitute for the institutional feedback that courts and regulators cannot deliver in time. Firms drifting toward narrative inflation are not lying — they are optimizing inside a degraded architecture, which is exactly why the fix is architectural rather than moral.

Lever 1 — Disciplined Validation Cadence

Closes detection lag. Defends the Product and Market dimensions.

Quarterly earnings concentrate validation into cliff events. The corrective is not an unbroken stream of market communications — over-disclosure destabilizes metrics, invites selective-communication risk, and runs into Regulation FD‘s constraints on selective disclosure of material nonpublic information. The corrective is a two-clock system: continuous internal validation on standardized metrics — adoption, utilization, AI revenue, productivity gains — registered externally on a periodic, uniform cadence. The internal clock catches the economics changing; the external clock publishes on schedule, to everyone at once, in stable units. Markets penalize uncertainty more than credible disappointment: a firm reporting flattening utilization with the data to prove it reprices less than a firm whose silence lets the market imagine worse. June’s evidence: the cohort members punished hardest were those whose economics arrived as revelation.

Lever 2 — Prospective Governance

Closes escalation lag. Defends the Governance dimension. Direct inheritance from The Duty to Foresee — AI Deployment Readiness as Prospective Governance, and the Arrival of Agentic Duty of Care, hereafter Agentic Duty of Care.

Governance escapes the compliance frame when it asks one forward question: what future disclosure could reasonably surprise investors? Then it answers that disclosure before the market forces the answer. Securities law does not require every forward-looking statement — but the duty to foresee, applied to capital-allocation representations, converts surprise-hunting from legal defense into management practice. The Hand-test arithmetic from AI Repricing Cycle applies: the burden of asking costs a rounding error against roughly $725 billion in planned annual deployment; the realized losses from not asking now include a 27 percent monthly drawdown and two live securities dockets.

Lever 3 — Dynamic Capital Allocation

Closes adaptation lag — and carries the framework’s sharpest tension.

Replace “we are spending $100 billion” with “spending scales with measurable demand milestones,” and capital re-registers automatically when assumptions change, instead of standing as a fixed representation the firm must later walk back. The tension: AI Repricing Cycle‘s Nash analysis showed no hyperscaler can unilaterally cut capex without signaling structural surrender. Milestone-gated spending risks reading as exactly that signal — unless the narrative converts conditionality into discipline. “We gate because we measure” is a strength claim; “we are reviewing our commitments” is a retreat claim, and markets price the difference. Meta’s June reversal proves mid-cycle reframing works. The deeper prize belongs to the first mover: one firm successfully reframing conditional capex as measurement discipline creates the focal point that gives the entire cohort permission to exit the equilibrium trap. The Dual Nash-Stigler Equilibrium Architecturenames both exits: a new Nash focal point among the players, or a Stigler-side institutional reset imposed from outside. The first mover chooses which one the industry gets.

Lever 4 — Internal Cognitive Digital Twins

Addresses a cross-system vulnerability — stakeholder anticipation. Defends the Infrastructure and Governance dimensions.

Firms simulate customers constantly; almost none simulate their own stakeholders. Model investors, regulators, competitors, plaintiffs, analysts, and rating agencies as interacting agents, then ask the only question that matters: which narrative breaks first? A simulation that repeatedly flags the same vulnerability — a demand representation outrunning capacity telemetry, a financing structure invisible to the balance sheet — hands management the corrective-disclosure candidate before external stakeholders find it. MindCast’s proprietary foresight simulations, disclosed at executive level in AI Repricing Cycle‘s appendix, run exactly this exercise from the outside; the lever internalizes it.

Lever 5 — Narrative Synchronization

Closes communication lag. Defends the Market and Governance dimensions.

Markets price coherence. A CEO announcing exploding demand, a CFO reporting flattening utilization, and a sales organization signaling slower enterprise adoption may all be individually accurate — and collectively incoherent, which markets treat as concealment even when it is merely disorganization. Incoherence of that kind is a coordination failure, not a character failure: the executives lack a shared focal point for what the AI story is, and focal points are infrastructure that must be built before the earnings call needs them — the Coase extension’s core finding, proved by an $80 billion enterprise that once spent five days unable to coordinate despite zero transaction costs. Cross-forum contradiction is precisely the pattern securities discovery mines — the audit MindCast’s AI Accountability: When AI Promises Meet the Courts series documents across four live dockets: earnings calls checked against filings, filings against sales decks, decks against executive interviews. The audit is mechanical — inventory every forum, score the deltas, close the largest gap first. Litigation runs the same audit later, with subpoena power.

Lever 6 — The AI Validation Dashboard

Addresses a cross-system vulnerability — observation asymmetry — and prices the framework’s central tradeoff.

Imagine a hyperscaler publishing quarterly: GPU utilization, Copilot-class adoption, enterprise ROI, capex efficiency, free-cash-flow trajectory, each with a trend arrow. Investors literally watch validation happening; the uncertainty premium shrinks; the validation appointment becomes a formality. The proportionality from the Executive Summary now earns its third term: the dashboard cuts registration lag and expectation sensitivity while enlarging representation surface — every published metric is a statement a restatement can contradict, a quietly redefined denominator becomes a pattern, and the discoverable-knowledge problem from Agentic Duty of Care applies in full. The exchange nets positive only under disciplined, consistent methodology: published definitions, stable baselines, no mid-stream redefinition. Transparency instruments reward exactly the firms that need them least, which is why deploying one is itself a credible signal.

III. The Owner: Organizational Foresight

Levers need an operator. Section II closed six distinct delays; the map’s final row names the failure no single lever fixes — a control system without an owner — and existing corporate functions do not fill the seat. Enterprise Risk Management looks backward at loss events; Internal Audit verifies controls; Compliance checks rules; Investor Relations transmits messages. No standard corporate function consistently owns the integrated question: what will investors believe six months from now if today’s trends continue — and what does the answer require of capital, disclosure, and governance now? Corporate strategy and foresight teams may ask the first half; connecting the answer to capital allocation, disclosure posture, litigation exposure, and market expectations is the seat that sits empty.

Organizational Foresight is that function — anticipation, not prediction. The capability operates each lever: it runs the stakeholder simulations (Lever 4), schedules the surprise audit (Lever 2), scores cross-forum coherence (Lever 5), sets the dashboard’s metrics before the market demands them (Lever 6), times the milestone narrative (Lever 3), and decides what the validation cadence registers next (Lever 1). Firms staff every other gap in the control architecture; the gap between internal reality and future external belief remains unstaffed at precisely the moment AI capital commitments made it the most expensive gap on the balance sheet.

Organizational Foresight arrives with an instrument panel, courtesy of the AGE framework: a pressure-to-control ratio and a governance-debt accumulator, both leading indicators that climb before any visible incident — the internal analog of the crossing calendar, a firm’s private validation appointment visible before the market books the public one. Formulas, and the structural argument for why the function cannot be automated away, appear in the Foundations appendix. One simulation finding sharpens the function’s reporting line: boards often recognize that marginal returns have changed before they feel authorized to slow investment, because prior approvals, competitor spending, and executive identity convert a financial question into a status-and-commitment problem. Foresight must reach the board early enough that an updated forecast reads as disciplined governance rather than repudiation of its own earlier decision.